Efficient Visualization of Whole Slide Images in Web-based Viewers for Digital Pathology.

Peter J. Schüffler, Evangelos Stamelos, Ishtiaque Ahmed, D. Vijay K. Yarlagadda, Matthew G. Hanna, Victor E. Reuter, David S. Klimstra and Meera Hameed.

Archives of Pathology & Laboratory Medicine, 2022-01-3

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_efficient_2022,
    title = {Efficient {Visualization} of {Whole} {Slide} {Images} in {Web}-based {Viewers} for {Digital} {Pathology}},
    issn = {1543-2165, 0003-9985},
    url = {https://meridian.allenpress.com/aplm/article/doi/10.5858/arpa.2021-0197-OA/476350/Efficient-Visualization-of-Whole-Slide-Images-in},
    doi = {10.5858/arpa.2021-0197-OA},
    abstract = {Context.—
     Wide adoption of digital pathology requires efficient visualization and navigation in Web-based digital slide viewers, which is poorly defined.


     Objective.—
     To define and quantify relevant performance metrics for efficient visualization of cases and slides in digital slide viewers.


     Design.—
     With a universal slide viewer used in clinical routine diagnostics, we evaluate the impact of slide caching, compression type, tile, and block size of whole slide images generated from Philips, Leica, and 3DHistech scanners on streaming performance on case, slide, and field of view levels.


     Results.—
     Two hundred thirty-nine pathologists routinely reviewed 60 080 whole slide images over 3 months. The median time to open a case's slides from the laboratory information system was less than 4 seconds, the time to change to a slide within the case was less than 1 second, and the time to render the adjacent field of view when navigating the slide was less than one-quarter of a second. A whole slide image's block size and a viewer tile size of 1024 pixels showed best performance to display a field of view and was preferrable over smaller tiles due to fewer mosaic effects. For Philips, fastest median slide streaming pace was 238 ms per field of view and for 3DHistech, 125 ms. For Leica, the fastest pace of 108 ms per field of view was established with block serving without decompression.


     Conclusions.—
     This is the first study to systematically assess user-centric slide visualization performance metrics for digital viewers, including time to open a case, time to change a slide, and time to change a field of view. These metrics help to improve the viewer's configuration, leading to an efficient visualization baseline that is widely accepted among pathologists using routine digital pathology.},
    language = {en},
    urldate = {2022-01-05},
    journal = {Archives of Pathology \& Laboratory Medicine},
    author = {Sch\"uffler, Peter J. and Stamelos, Evangelos and Ahmed, Ishtiaque and Yarlagadda, D. Vijay K. and Hanna, Matthew G. and Reuter, Victor E. and Klimstra, David S. and Hameed, Meera},
    month = jan,
    year = {2022},
}

Download Endnote/RIS citation

TY - JOUR
TI - Efficient Visualization of Whole Slide Images in Web-based Viewers for Digital Pathology
AU - Schüffler, Peter J.
AU - Stamelos, Evangelos
AU - Ahmed, Ishtiaque
AU - Yarlagadda, D. Vijay K.
AU - Hanna, Matthew G.
AU - Reuter, Victor E.
AU - Klimstra, David S.
AU - Hameed, Meera
T2 - Archives of Pathology & Laboratory Medicine
AB - Context.—
Wide adoption of digital pathology requires efficient visualization and navigation in Web-based digital slide viewers, which is poorly defined.

Objective.—
To define and quantify relevant performance metrics for efficient visualization of cases and slides in digital slide viewers.

Design.—
With a universal slide viewer used in clinical routine diagnostics, we evaluate the impact of slide caching, compression type, tile, and block size of whole slide images generated from Philips, Leica, and 3DHistech scanners on streaming performance on case, slide, and field of view levels.

Results.—
Two hundred thirty-nine pathologists routinely reviewed 60 080 whole slide images over 3 months. The median time to open a case's slides from the laboratory information system was less than 4 seconds, the time to change to a slide within the case was less than 1 second, and the time to render the adjacent field of view when navigating the slide was less than one-quarter of a second. A whole slide image's block size and a viewer tile size of 1024 pixels showed best performance to display a field of view and was preferrable over smaller tiles due to fewer mosaic effects. For Philips, fastest median slide streaming pace was 238 ms per field of view and for 3DHistech, 125 ms. For Leica, the fastest pace of 108 ms per field of view was established with block serving without decompression.

Conclusions.—
This is the first study to systematically assess user-centric slide visualization performance metrics for digital viewers, including time to open a case, time to change a slide, and time to change a field of view. These metrics help to improve the viewer's configuration, leading to an efficient visualization baseline that is widely accepted among pathologists using routine digital pathology.
DA - 2022/01/03/
PY - 2022
DO - 10.5858/arpa.2021-0197-OA
DP - DOI.org (Crossref)
LA - en
SN - 1543-2165, 0003-9985
UR - https://meridian.allenpress.com/aplm/article/doi/10.5858/arpa.2021-0197-OA/476350/Efficient-Visualization-of-Whole-Slide-Images-in
Y2 - 2022/01/05/09:25:21
ER -

Context.— Wide adoption of digital pathology requires efficient visualization and navigation in Web-based digital slide viewers, which is poorly defined. Objective.— To define and quantify relevant performance metrics for efficient visualization of cases and slides in digital slide viewers. Design.— With a universal slide viewer used in clinical routine diagnostics, we evaluate the impact of slide caching, compression type, tile, and block size of whole slide images generated from Philips, Leica, and 3DHistech scanners on streaming performance on case, slide, and field of view levels. Results.— Two hundred thirty-nine pathologists routinely reviewed 60 080 whole slide images over 3 months. The median time to open a case's slides from the laboratory information system was less than 4 seconds, the time to change to a slide within the case was less than 1 second, and the time to render the adjacent field of view when navigating the slide was less than one-quarter of a second. A whole slide image's block size and a viewer tile size of 1024 pixels showed best performance to display a field of view and was preferrable over smaller tiles due to fewer mosaic effects. For Philips, fastest median slide streaming pace was 238 ms per field of view and for 3DHistech, 125 ms. For Leica, the fastest pace of 108 ms per field of view was established with block serving without decompression. Conclusions.— This is the first study to systematically assess user-centric slide visualization performance metrics for digital viewers, including time to open a case, time to change a slide, and time to change a field of view. These metrics help to improve the viewer's configuration, leading to an efficient visualization baseline that is widely accepted among pathologists using routine digital pathology.

Integrated digital pathology at scale: A solution for clinical diagnostics and cancer research at a large academic medical center.

Peter J Schüffler, Luke Geneslaw, D Vijay K Yarlagadda, Matthew G Hanna, Jennifer Samboy, Evangelos Stamelos, Chad Vanderbilt, John Philip, Marc-Henri Jean, Lorraine Corsale, Allyne Manzo, Neeraj H G Paramasivam, John S Ziegler, Jianjiong Gao, Juan C Perin, Young Suk Kim, Umeshkumar K Bhanot, Michael H A Roehrl, Orly Ardon, Sarah Chiang, Dilip D Giri, Carlie S Sigel, Lee K Tan, Melissa Murray, Christina Virgo, Christine England, Yukako Yagi, S Joseph Sirintrapun, David Klimstra, Meera Hameed, Victor E Reuter and Thomas J Fuchs.

Journal of the American Medical Informatics Association, vol. 28, 9, p. 1874-1884, July 14, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_integrated_2021,
    title = {Integrated digital pathology at scale: {A} solution for clinical diagnostics and cancer research at a large academic medical center},
    volume = {28},
    issn = {1527-974X},
    shorttitle = {Integrated digital pathology at scale},
    url = {https://doi.org/10.1093/jamia/ocab085},
    doi = {10.1093/jamia/ocab085},
    abstract = {Broad adoption of digital pathology (DP) is still lacking, and examples for DP connecting diagnostic, research, and educational use cases are missing. We blueprint a holistic DP solution at a large academic medical center ubiquitously integrated into clinical workflows; researchapplications including molecular, genetic, and tissue databases; and educational processes.We built a vendor-agnostic, integrated viewer for reviewing, annotating, sharing, and quality assurance of digital slides in a clinical or research context. It is the first homegrown viewer cleared by New York State provisional approval in 2020 for primary diagnosis and remote sign-out during the COVID-19 (coronavirus disease 2019) pandemic. We further introduce an interconnected Honest Broker for BioInformatics Technology (HoBBIT) to systematically compile and share large-scale DP research datasets including anonymized images, redacted pathology reports, and clinical data of patients with consent.The solution has been operationally used over 3 years by 926 pathologists and researchers evaluating 288 903 digital slides. A total of 51\% of these were reviewed within 1 month after scanning. Seamless integration of the viewer into 4 hospital systems clearly increases the adoption of DP. HoBBIT directly impacts the translation of knowledge in pathology into effective new health measures, including artificial intelligence–driven detection models for prostate cancer, basal cell carcinoma, and breast cancer metastases, developed and validated on thousands of cases.We highlight major challenges and lessons learned when going digital to provide orientation for other pathologists. Building interconnected solutions will not only increase adoption of DP, but also facilitate next-generation computational pathology at scale for enhanced cancer research.},
    number = {9},
    urldate = {2021-07-14},
    journal = {Journal of the American Medical Informatics Association},
    author = {Sch\"uffler, Peter J and Geneslaw, Luke and Yarlagadda, D Vijay K and Hanna, Matthew G and Samboy, Jennifer and Stamelos, Evangelos and Vanderbilt, Chad and Philip, John and Jean, Marc-Henri and Corsale, Lorraine and Manzo, Allyne and Paramasivam, Neeraj H G and Ziegler, John S and Gao, Jianjiong and Perin, Juan C and Kim, Young Suk and Bhanot, Umeshkumar K and Roehrl, Michael H A and Ardon, Orly and Chiang, Sarah and Giri, Dilip D and Sigel, Carlie S and Tan, Lee K and Murray, Melissa and Virgo, Christina and England, Christine and Yagi, Yukako and Sirintrapun, S Joseph and Klimstra, David and Hameed, Meera and Reuter, Victor E and Fuchs, Thomas J},
    month = jul,
    year = {2021},
    pages = {1874--1884},
}

Download Endnote/RIS citation

TY - JOUR
TI - Integrated digital pathology at scale: A solution for clinical diagnostics and cancer research at a large academic medical center
AU - Schüffler, Peter J
AU - Geneslaw, Luke
AU - Yarlagadda, D Vijay K
AU - Hanna, Matthew G
AU - Samboy, Jennifer
AU - Stamelos, Evangelos
AU - Vanderbilt, Chad
AU - Philip, John
AU - Jean, Marc-Henri
AU - Corsale, Lorraine
AU - Manzo, Allyne
AU - Paramasivam, Neeraj H G
AU - Ziegler, John S
AU - Gao, Jianjiong
AU - Perin, Juan C
AU - Kim, Young Suk
AU - Bhanot, Umeshkumar K
AU - Roehrl, Michael H A
AU - Ardon, Orly
AU - Chiang, Sarah
AU - Giri, Dilip D
AU - Sigel, Carlie S
AU - Tan, Lee K
AU - Murray, Melissa
AU - Virgo, Christina
AU - England, Christine
AU - Yagi, Yukako
AU - Sirintrapun, S Joseph
AU - Klimstra, David
AU - Hameed, Meera
AU - Reuter, Victor E
AU - Fuchs, Thomas J
T2 - Journal of the American Medical Informatics Association
AB - Broad adoption of digital pathology (DP) is still lacking, and examples for DP connecting diagnostic, research, and educational use cases are missing. We blueprint a holistic DP solution at a large academic medical center ubiquitously integrated into clinical workflows; researchapplications including molecular, genetic, and tissue databases; and educational processes.We built a vendor-agnostic, integrated viewer for reviewing, annotating, sharing, and quality assurance of digital slides in a clinical or research context. It is the first homegrown viewer cleared by New York State provisional approval in 2020 for primary diagnosis and remote sign-out during the COVID-19 (coronavirus disease 2019) pandemic. We further introduce an interconnected Honest Broker for BioInformatics Technology (HoBBIT) to systematically compile and share large-scale DP research datasets including anonymized images, redacted pathology reports, and clinical data of patients with consent.The solution has been operationally used over 3 years by 926 pathologists and researchers evaluating 288 903 digital slides. A total of 51% of these were reviewed within 1 month after scanning. Seamless integration of the viewer into 4 hospital systems clearly increases the adoption of DP. HoBBIT directly impacts the translation of knowledge in pathology into effective new health measures, including artificial intelligence–driven detection models for prostate cancer, basal cell carcinoma, and breast cancer metastases, developed and validated on thousands of cases.We highlight major challenges and lessons learned when going digital to provide orientation for other pathologists. Building interconnected solutions will not only increase adoption of DP, but also facilitate next-generation computational pathology at scale for enhanced cancer research.
DA - 2021/07/14/
PY - 2021
DO - 10.1093/jamia/ocab085
DP - Silverchair
VL - 28
IS - 9
SP - 1874
EP - 1884
J2 - Journal of the American Medical Informatics Association
SN - 1527-974X
ST - Integrated digital pathology at scale
UR - https://doi.org/10.1093/jamia/ocab085
Y2 - 2021/07/14/21:13:00
ER -

Broad adoption of digital pathology (DP) is still lacking, and examples for DP connecting diagnostic, research, and educational use cases are missing. We blueprint a holistic DP solution at a large academic medical center ubiquitously integrated into clinical workflows; researchapplications including molecular, genetic, and tissue databases; and educational processes.We built a vendor-agnostic, integrated viewer for reviewing, annotating, sharing, and quality assurance of digital slides in a clinical or research context. It is the first homegrown viewer cleared by New York State provisional approval in 2020 for primary diagnosis and remote sign-out during the COVID-19 (coronavirus disease 2019) pandemic. We further introduce an interconnected Honest Broker for BioInformatics Technology (HoBBIT) to systematically compile and share large-scale DP research datasets including anonymized images, redacted pathology reports, and clinical data of patients with consent.The solution has been operationally used over 3 years by 926 pathologists and researchers evaluating 288 903 digital slides. A total of 51% of these were reviewed within 1 month after scanning. Seamless integration of the viewer into 4 hospital systems clearly increases the adoption of DP. HoBBIT directly impacts the translation of knowledge in pathology into effective new health measures, including artificial intelligence–driven detection models for prostate cancer, basal cell carcinoma, and breast cancer metastases, developed and validated on thousands of cases.We highlight major challenges and lessons learned when going digital to provide orientation for other pathologists. Building interconnected solutions will not only increase adoption of DP, but also facilitate next-generation computational pathology at scale for enhanced cancer research.

Digital Pathology Operations at an NYC Tertiary Cancer Center During the First 4 Months of COVID-19 Pandemic Response.

Orly Ardon, Victor E. Reuter, Meera Hameed, Lorraine Corsale, Allyne Manzo, Sahussapont J. Sirintrapun, Peter Ntiamoah, Evangelos Stamelos, Peter J. Schueffler, Christine England, David S. Klimstra and Matthew G. Hanna.

Academic Pathology, vol. 8, p. 23742895211010276, April 28, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{ardon_digital_2021,
    title = {Digital {Pathology} {Operations} at an {NYC} {Tertiary} {Cancer} {Center} {During} the {First} 4 {Months} of {COVID}-19 {Pandemic} {Response}},
    volume = {8},
    issn = {2374-2895},
    url = {https://doi.org/10.1177/23742895211010276},
    doi = {10.1177/23742895211010276},
    abstract = {Implementation of an infrastructure to support digital pathology began in 2006 at Memorial Sloan Kettering Cancer Center. The public health emergency and COVID-19 pandemic regulations in New York City required a novel workflow to sustain existing operations. While regulatory enforcement discretions offered faculty workspace flexibility, a substantial portion of laboratory and digital pathology workflows require on-site presence of staff. Maintaining social distancing and offering staggered work schedules. Due to a decrease in patients seeking health care at the onset of the pandemic, a temporary decrease in patient specimens was observed. Hospital and travel regulations impacted onsite vendor technical support. Digital glass slide scanning activities onsite proceeded without interruption throughout the pandemic, with challenges including staff who required quarantine due to virus exposure, unrelated illness, family support, or lack of public transportation. During the public health emergency, we validated digital pathology systems for a remote pathology operation. Since March 2020, the departmental digital pathology staff were able to maintain scanning volumes of over 100 000 slides per month. The digital scanning team reprioritized archival slide scanning and participated in a remote sign-out validation and successful submission of New York State approval for a laboratory developed test. Digital pathology offers a health care delivery model where pathologists can perform their sign out duties at remote location and prevent disruptions to critical pathology services for patients seeking care at our institution during emergencies. Development of standard operating procedures to support digital workflows will maintain turnaround times and enable clinical operations during emergency or otherwise unanticipated events.},
    language = {en},
    urldate = {2021-09-01},
    journal = {Academic Pathology},
    author = {Ardon, Orly and Reuter, Victor E. and Hameed, Meera and Corsale, Lorraine and Manzo, Allyne and Sirintrapun, Sahussapont J. and Ntiamoah, Peter and Stamelos, Evangelos and Schueffler, Peter J. and England, Christine and Klimstra, David S. and Hanna, Matthew G.},
    month = apr,
    year = {2021},
    keywords = {COVID-19, clinical, digital pathology, implementation, operations, remote signout, telepathology},
    pages = {23742895211010276},
}

Download Endnote/RIS citation

TY - JOUR
TI - Digital Pathology Operations at an NYC Tertiary Cancer Center During the First 4 Months of COVID-19 Pandemic Response
AU - Ardon, Orly
AU - Reuter, Victor E.
AU - Hameed, Meera
AU - Corsale, Lorraine
AU - Manzo, Allyne
AU - Sirintrapun, Sahussapont J.
AU - Ntiamoah, Peter
AU - Stamelos, Evangelos
AU - Schueffler, Peter J.
AU - England, Christine
AU - Klimstra, David S.
AU - Hanna, Matthew G.
T2 - Academic Pathology
AB - Implementation of an infrastructure to support digital pathology began in 2006 at Memorial Sloan Kettering Cancer Center. The public health emergency and COVID-19 pandemic regulations in New York City required a novel workflow to sustain existing operations. While regulatory enforcement discretions offered faculty workspace flexibility, a substantial portion of laboratory and digital pathology workflows require on-site presence of staff. Maintaining social distancing and offering staggered work schedules. Due to a decrease in patients seeking health care at the onset of the pandemic, a temporary decrease in patient specimens was observed. Hospital and travel regulations impacted onsite vendor technical support. Digital glass slide scanning activities onsite proceeded without interruption throughout the pandemic, with challenges including staff who required quarantine due to virus exposure, unrelated illness, family support, or lack of public transportation. During the public health emergency, we validated digital pathology systems for a remote pathology operation. Since March 2020, the departmental digital pathology staff were able to maintain scanning volumes of over 100 000 slides per month. The digital scanning team reprioritized archival slide scanning and participated in a remote sign-out validation and successful submission of New York State approval for a laboratory developed test. Digital pathology offers a health care delivery model where pathologists can perform their sign out duties at remote location and prevent disruptions to critical pathology services for patients seeking care at our institution during emergencies. Development of standard operating procedures to support digital workflows will maintain turnaround times and enable clinical operations during emergency or otherwise unanticipated events.
DA - 2021/04/28/
PY - 2021
DO - 10.1177/23742895211010276
DP - SAGE Journals
VL - 8
SP - 23742895211010276
J2 - Academic Pathology
LA - en
SN - 2374-2895
UR - https://doi.org/10.1177/23742895211010276
Y2 - 2021/09/01/07:50:28
KW - COVID-19
KW - clinical
KW - digital pathology
KW - implementation
KW - operations
KW - remote signout
KW - telepathology
ER -

Implementation of an infrastructure to support digital pathology began in 2006 at Memorial Sloan Kettering Cancer Center. The public health emergency and COVID-19 pandemic regulations in New York City required a novel workflow to sustain existing operations. While regulatory enforcement discretions offered faculty workspace flexibility, a substantial portion of laboratory and digital pathology workflows require on-site presence of staff. Maintaining social distancing and offering staggered work schedules. Due to a decrease in patients seeking health care at the onset of the pandemic, a temporary decrease in patient specimens was observed. Hospital and travel regulations impacted onsite vendor technical support. Digital glass slide scanning activities onsite proceeded without interruption throughout the pandemic, with challenges including staff who required quarantine due to virus exposure, unrelated illness, family support, or lack of public transportation. During the public health emergency, we validated digital pathology systems for a remote pathology operation. Since March 2020, the departmental digital pathology staff were able to maintain scanning volumes of over 100 000 slides per month. The digital scanning team reprioritized archival slide scanning and participated in a remote sign-out validation and successful submission of New York State approval for a laboratory developed test. Digital pathology offers a health care delivery model where pathologists can perform their sign out duties at remote location and prevent disruptions to critical pathology services for patients seeking care at our institution during emergencies. Development of standard operating procedures to support digital workflows will maintain turnaround times and enable clinical operations during emergency or otherwise unanticipated events.

Multi-magnification-based machine learning as an ancillary tool for the pathologic assessment of shaved margins for breast carcinoma lumpectomy specimens.

Timothy M. D’Alfonso, David Joon Ho, Matthew G. Hanna, Anne Grabenstetter, Dig Vijay Kumar Yarlagadda, Luke Geneslaw, Peter Ntiamoah, Thomas J. Fuchs and Lee K. Tan.

Modern Pathology, 4-26-2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{dalfonso_multi-magnification-based_2021,
    title = {Multi-magnification-based machine learning as an ancillary tool for the pathologic assessment of shaved margins for breast carcinoma lumpectomy specimens},
    url = {https://doi.org/10.1038/s41379-021-00807-9},
    doi = {10.1038/s41379-021-00807-9},
    abstract = {The surgical margin status of breast lumpectomy specimens for invasive carcinoma and ductal carcinoma in situ (DCIS) guides clinical decisions, as positive margins are associated with higher rates of local recurrence. The “cavity shave” method of margin assessment has the benefits of allowing the surgeon to orient shaved margins intraoperatively and the pathologist to assess one inked margin per specimen. We studied whether a deep convolutional neural network, a deep multi-magnification network (DMMN), could accurately segment carcinoma from benign tissue in whole slide images (WSIs) of shave margin slides, and therefore serve as a potential screening tool to improve the efficiency of microscopic evaluation of these specimens. Applying the pretrained DMMN model, or the initial model, to a validation set of 408 WSIs (348 benign, 60 with carcinoma) achieved an area under the curve (AUC) of 0.941. After additional manual annotations and fine-tuning of the model, the updated model achieved an AUC of 0.968 with sensitivity set at 100\% and corresponding specificity of 78\%. We applied the initial model and updated model to a testing set of 427 WSIs (374 benign, 53 with carcinoma) which showed AUC values of 0.900 and 0.927, respectively. Using the pixel classification threshold selected from the validation set, the model achieved a sensitivity of 92\% and specificity of 78\%. The four false-negative classifications resulted from two small foci of DCIS (1 mm, 0.5 mm) and two foci of well-differentiated invasive carcinoma (3 mm, 1.5 mm). This proof-of-principle study demonstrates that a DMMN machine learning model can segment invasive carcinoma and DCIS in surgical margin specimens with high accuracy and has the potential to be used as a screening tool for pathologic assessment of these specimens.},
    journal = {Modern Pathology},
    author = {D’Alfonso, Timothy M. and Ho, David Joon and Hanna, Matthew G. and Grabenstetter, Anne and Yarlagadda, Dig Vijay Kumar and Geneslaw, Luke and Ntiamoah, Peter and Fuchs, Thomas J. and Tan, Lee K.},
    month = apr,
    year = {2021},
}

Download Endnote/RIS citation

TY - JOUR
TI - Multi-magnification-based machine learning as an ancillary tool for the pathologic assessment of shaved margins for breast carcinoma lumpectomy specimens
AU - D’Alfonso, Timothy M.
AU - Ho, David Joon
AU - Hanna, Matthew G.
AU - Grabenstetter, Anne
AU - Yarlagadda, Dig Vijay Kumar
AU - Geneslaw, Luke
AU - Ntiamoah, Peter
AU - Fuchs, Thomas J.
AU - Tan, Lee K.
T2 - Modern Pathology
AB - The surgical margin status of breast lumpectomy specimens for invasive carcinoma and ductal carcinoma in situ (DCIS) guides clinical decisions, as positive margins are associated with higher rates of local recurrence. The “cavity shave” method of margin assessment has the benefits of allowing the surgeon to orient shaved margins intraoperatively and the pathologist to assess one inked margin per specimen. We studied whether a deep convolutional neural network, a deep multi-magnification network (DMMN), could accurately segment carcinoma from benign tissue in whole slide images (WSIs) of shave margin slides, and therefore serve as a potential screening tool to improve the efficiency of microscopic evaluation of these specimens. Applying the pretrained DMMN model, or the initial model, to a validation set of 408 WSIs (348 benign, 60 with carcinoma) achieved an area under the curve (AUC) of 0.941. After additional manual annotations and fine-tuning of the model, the updated model achieved an AUC of 0.968 with sensitivity set at 100% and corresponding specificity of 78%. We applied the initial model and updated model to a testing set of 427 WSIs (374 benign, 53 with carcinoma) which showed AUC values of 0.900 and 0.927, respectively. Using the pixel classification threshold selected from the validation set, the model achieved a sensitivity of 92% and specificity of 78%. The four false-negative classifications resulted from two small foci of DCIS (1 mm, 0.5 mm) and two foci of well-differentiated invasive carcinoma (3 mm, 1.5 mm). This proof-of-principle study demonstrates that a DMMN machine learning model can segment invasive carcinoma and DCIS in surgical margin specimens with high accuracy and has the potential to be used as a screening tool for pathologic assessment of these specimens.
DA - 2021/04/26/
PY - 2021
DO - 10.1038/s41379-021-00807-9
UR - https://doi.org/10.1038/s41379-021-00807-9
ER -

The surgical margin status of breast lumpectomy specimens for invasive carcinoma and ductal carcinoma in situ (DCIS) guides clinical decisions, as positive margins are associated with higher rates of local recurrence. The “cavity shave” method of margin assessment has the benefits of allowing the surgeon to orient shaved margins intraoperatively and the pathologist to assess one inked margin per specimen. We studied whether a deep convolutional neural network, a deep multi-magnification network (DMMN), could accurately segment carcinoma from benign tissue in whole slide images (WSIs) of shave margin slides, and therefore serve as a potential screening tool to improve the efficiency of microscopic evaluation of these specimens. Applying the pretrained DMMN model, or the initial model, to a validation set of 408 WSIs (348 benign, 60 with carcinoma) achieved an area under the curve (AUC) of 0.941. After additional manual annotations and fine-tuning of the model, the updated model achieved an AUC of 0.968 with sensitivity set at 100% and corresponding specificity of 78%. We applied the initial model and updated model to a testing set of 427 WSIs (374 benign, 53 with carcinoma) which showed AUC values of 0.900 and 0.927, respectively. Using the pixel classification threshold selected from the validation set, the model achieved a sensitivity of 92% and specificity of 78%. The four false-negative classifications resulted from two small foci of DCIS (1 mm, 0.5 mm) and two foci of well-differentiated invasive carcinoma (3 mm, 1.5 mm). This proof-of-principle study demonstrates that a DMMN machine learning model can segment invasive carcinoma and DCIS in surgical margin specimens with high accuracy and has the potential to be used as a screening tool for pathologic assessment of these specimens.

Overcoming an annotation hurdle: Digitizing pen annotations from whole slide images.

Peter J. Schüffler, Dig Vijay Kumar Yarlagadda, Chad Vanderbilt and Thomas J. Fuchs.

Journal of Pathology Informatics, vol. 12, 1, p. 9, 02/23/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_overcoming_2021,
    title = {Overcoming an annotation hurdle: {Digitizing} pen annotations from whole slide images},
    volume = {12},
    issn = {2153-3539},
    shorttitle = {Overcoming an annotation hurdle},
    url = {https://www.doi.org/10.4103/jpi.jpi_85_20},
    doi = {10.4103/jpi.jpi_85_20},
    abstract = {Background:The development of artificial intelligence (AI) in pathology frequently relies on digitally annotated whole slide images (WSI). The creation of these annotations – manually drawn by pathologists in digital slide viewers – is time consuming and expensive. At the same time, pathologists routinely annotate glass slides with a pen to outline cancerous regions, for example, for molecular assessment of the tissue. These pen annotations are currently considered artifacts and excluded from computational modeling. Methods:We propose a novel method to segment and fill hand-drawn pen annotations and convert them into a digital format to make them accessible for computational models. Our method is implemented in Python as an open source, publicly available software tool. Results: Our method is able to extract pen annotations from WSI and save them as annotation masks. On a data set of 319 WSI with pen markers, we validate our algorithm segmenting the annotations with an overall Dice metric of 0.942, Precision of 0.955, and Recall of 0.943. Processing all images takes 15 min in contrast to 5 h manual digital annotation time. Further, the approach is robust against text annotations. Conclusions: We envision that our method can take advantage of already pen-annotated slides in scenarios in which the annotations would be helpful for training computational models. We conclude that, considering the large archives of many pathology departments that are currently being digitized, our method will help to collect large numbers of training samples from those data.},
    language = {en},
    number = {1},
    urldate = {2021-02-25},
    journal = {Journal of Pathology Informatics},
    author = {Sch\"uffler, Peter J. and Yarlagadda, Dig Vijay Kumar and Vanderbilt, Chad and Fuchs, Thomas J.},
    month = feb,
    year = {2021},
    Distributor: Medknow Publications and Media Pvt. Ltd.
    Institution: Medknow Publications and Media Pvt. Ltd.
    Label: Medknow Publications and Media Pvt. Ltd.
    Publisher: Medknow Publications},
    pages = {9},
}

Download Endnote/RIS citation

TY - JOUR
TI - Overcoming an annotation hurdle: Digitizing pen annotations from whole slide images
AU - Schüffler, Peter J.
AU - Yarlagadda, Dig Vijay Kumar
AU - Vanderbilt, Chad
AU - Fuchs, Thomas J.
T2 - Journal of Pathology Informatics
AB - Background:The development of artificial intelligence (AI) in pathology frequently relies on digitally annotated whole slide images (WSI). The creation of these annotations – manually drawn by pathologists in digital slide viewers – is time consuming and expensive. At the same time, pathologists routinely annotate glass slides with a pen to outline cancerous regions, for example, for molecular assessment of the tissue. These pen annotations are currently considered artifacts and excluded from computational modeling. Methods:We propose a novel method to segment and fill hand-drawn pen annotations and convert them into a digital format to make them accessible for computational models. Our method is implemented in Python as an open source, publicly available software tool. Results: Our method is able to extract pen annotations from WSI and save them as annotation masks. On a data set of 319 WSI with pen markers, we validate our algorithm segmenting the annotations with an overall Dice metric of 0.942, Precision of 0.955, and Recall of 0.943. Processing all images takes 15 min in contrast to 5 h manual digital annotation time. Further, the approach is robust against text annotations. Conclusions: We envision that our method can take advantage of already pen-annotated slides in scenarios in which the annotations would be helpful for training computational models. We conclude that, considering the large archives of many pathology departments that are currently being digitized, our method will help to collect large numbers of training samples from those data.
DA - 2021/02/23/
PY - 2021
DO - 10.4103/jpi.jpi_85_20
DP - www.jpathinformatics.org
VL - 12
IS - 1
SP - 9
LA - en
SN - 2153-3539
ST - Overcoming an annotation hurdle
UR - https://www.doi.org/10.4103/jpi.jpi_85_20
Y2 - 2021/02/25/18:48:46
ER -

Background:The development of artificial intelligence (AI) in pathology frequently relies on digitally annotated whole slide images (WSI). The creation of these annotations – manually drawn by pathologists in digital slide viewers – is time consuming and expensive. At the same time, pathologists routinely annotate glass slides with a pen to outline cancerous regions, for example, for molecular assessment of the tissue. These pen annotations are currently considered artifacts and excluded from computational modeling. Methods:We propose a novel method to segment and fill hand-drawn pen annotations and convert them into a digital format to make them accessible for computational models. Our method is implemented in Python as an open source, publicly available software tool. Results: Our method is able to extract pen annotations from WSI and save them as annotation masks. On a data set of 319 WSI with pen markers, we validate our algorithm segmenting the annotations with an overall Dice metric of 0.942, Precision of 0.955, and Recall of 0.943. Processing all images takes 15 min in contrast to 5 h manual digital annotation time. Further, the approach is robust against text annotations. Conclusions: We envision that our method can take advantage of already pen-annotated slides in scenarios in which the annotations would be helpful for training computational models. We conclude that, considering the large archives of many pathology departments that are currently being digitized, our method will help to collect large numbers of training samples from those data.

Deep Multi-Magnification Networks for Multi-Class Breast Cancer Image Segmentation.

David Joon Ho, Dig V. K. Yarlagadda, Timothy M. D'Alfonso, Matthew G. Hanna, Anne Grabenstetter, Peter Ntiamoah, Edi Brogi, Lee K. Tan and Thomas J. Fuchs.

Computerized Medical Imaging and Graphics, vol. 88, 1/12/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{ho_deep_2021,
    title = {Deep {Multi}-{Magnification} {Networks} for {Multi}-{Class} {Breast} {Cancer} {Image} {Segmentation}},
    volume = {88},
    url = {https://doi.org/10.1016/j.compmedimag.2021.101866},
    doi = {10.1016/j.compmedimag.2021.101866},
    abstract = {Pathologic analysis of surgical excision specimens for breast carcinoma is important to evaluate the completeness of surgical excision and has implications for future treatment. This analysis is performed manually by pathologists reviewing histologic slides prepared from formalin-fixed tissue. In this paper, we present Deep Multi-Magnification Network trained by partial annotation for automated multi-class tissue segmentation by a set of patches from multiple magnifications in digitized whole slide images. Our proposed architecture with multi-encoder, multi-decoder, and multi-concatenation outperforms other single and multi-magnification-based architectures by achieving the highest mean intersection-over-union, and can be used to facilitate pathologists’ assessments of breast cancer.},
    urldate = {2019-11-14},
    journal = {Computerized Medical Imaging and Graphics},
    author = {Ho, David Joon and Yarlagadda, Dig V. K. and D'Alfonso, Timothy M. and Hanna, Matthew G. and Grabenstetter, Anne and Ntiamoah, Peter and Brogi, Edi and Tan, Lee K. and Fuchs, Thomas J.},
    month = jan,
    year = {2021},
    keywords = {Computer Science - Computer Vision and Pattern Recognition, Electrical Engineering and Systems Science - Image and Video Processing},
}

Download Endnote/RIS citation

TY - JOUR
TI - Deep Multi-Magnification Networks for Multi-Class Breast Cancer Image Segmentation
AU - Ho, David Joon
AU - Yarlagadda, Dig V. K.
AU - D'Alfonso, Timothy M.
AU - Hanna, Matthew G.
AU - Grabenstetter, Anne
AU - Ntiamoah, Peter
AU - Brogi, Edi
AU - Tan, Lee K.
AU - Fuchs, Thomas J.
T2 - Computerized Medical Imaging and Graphics
AB - Pathologic analysis of surgical excision specimens for breast carcinoma is important to evaluate the completeness of surgical excision and has implications for future treatment. This analysis is performed manually by pathologists reviewing histologic slides prepared from formalin-fixed tissue. In this paper, we present Deep Multi-Magnification Network trained by partial annotation for automated multi-class tissue segmentation by a set of patches from multiple magnifications in digitized whole slide images. Our proposed architecture with multi-encoder, multi-decoder, and multi-concatenation outperforms other single and multi-magnification-based architectures by achieving the highest mean intersection-over-union, and can be used to facilitate pathologists’ assessments of breast cancer.
DA - 2021/01/12/
PY - 2021
DO - 10.1016/j.compmedimag.2021.101866
VL - 88
UR - https://doi.org/10.1016/j.compmedimag.2021.101866
Y2 - 2019/11/14/16:41:13
KW - Computer Science - Computer Vision and Pattern Recognition
KW - Electrical Engineering and Systems Science - Image and Video Processing
ER -

Pathologic analysis of surgical excision specimens for breast carcinoma is important to evaluate the completeness of surgical excision and has implications for future treatment. This analysis is performed manually by pathologists reviewing histologic slides prepared from formalin-fixed tissue. In this paper, we present Deep Multi-Magnification Network trained by partial annotation for automated multi-class tissue segmentation by a set of patches from multiple magnifications in digitized whole slide images. Our proposed architecture with multi-encoder, multi-decoder, and multi-concatenation outperforms other single and multi-magnification-based architectures by achieving the highest mean intersection-over-union, and can be used to facilitate pathologists’ assessments of breast cancer.

Deep Learning Methods for Lung Cancer Segmentation in Whole-Slide Histopathology Images—The ACDC@LungHP Challenge 2019.

Zhang Li, Jiehua Zhang, Tao Tan, Xichao Teng, Xiaoliang Sun, Hong Zhao, Lihong Liu, Yang Xiao, Byungjae Lee, Yilong Li, Qianni Zhang, Shujiao Sun, Yushan Zheng, Junyu Yan, Ni Li, Yiyu Hong, Junsu Ko, Hyun Jung, Yanling Liu, Yu-cheng Chen, Ching-wei Wang, Vladimir Yurovskiy, Pavel Maevskikh, Vahid Khanagha, Yi Jiang, Li Yu, Zhihong Liu, Daiqiang Li, Peter J. Schuffler, Qifeng Yu, Hui Chen, Yuling Tang and Geert Litjens.

IEEE J. Biomed. Health Inform., vol. 25, 2, p. 429-440, 2/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{li_deep_2021,
    title = {Deep {Learning} {Methods} for {Lung} {Cancer} {Segmentation} in {Whole}-{Slide} {Histopathology} {Images}—{The} {ACDC}@{LungHP} {Challenge} 2019},
    volume = {25},
    issn = {2168-2194, 2168-2208},
    url = {https://ieeexplore.ieee.org/document/9265237/},
    doi = {10.1109/JBHI.2020.3039741},
    number = {2},
    urldate = {2022-07-13},
    journal = {IEEE Journal of Biomedical and Health Informatics},
    author = {Li, Zhang and Zhang, Jiehua and Tan, Tao and Teng, Xichao and Sun, Xiaoliang and Zhao, Hong and Liu, Lihong and Xiao, Yang and Lee, Byungjae and Li, Yilong and Zhang, Qianni and Sun, Shujiao and Zheng, Yushan and Yan, Junyu and Li, Ni and Hong, Yiyu and Ko, Junsu and Jung, Hyun and Liu, Yanling and Chen, Yu-cheng and Wang, Ching-wei and Yurovskiy, Vladimir and Maevskikh, Pavel and Khanagha, Vahid and Jiang, Yi and Yu, Li and Liu, Zhihong and Li, Daiqiang and Schuffler, Peter J. and Yu, Qifeng and Chen, Hui and Tang, Yuling and Litjens, Geert},
    month = feb,
    year = {2021},
    pages = {429--440},
}

Download Endnote/RIS citation

TY - JOUR
TI - Deep Learning Methods for Lung Cancer Segmentation in Whole-Slide Histopathology Images—The ACDC@LungHP Challenge 2019
AU - Li, Zhang
AU - Zhang, Jiehua
AU - Tan, Tao
AU - Teng, Xichao
AU - Sun, Xiaoliang
AU - Zhao, Hong
AU - Liu, Lihong
AU - Xiao, Yang
AU - Lee, Byungjae
AU - Li, Yilong
AU - Zhang, Qianni
AU - Sun, Shujiao
AU - Zheng, Yushan
AU - Yan, Junyu
AU - Li, Ni
AU - Hong, Yiyu
AU - Ko, Junsu
AU - Jung, Hyun
AU - Liu, Yanling
AU - Chen, Yu-cheng
AU - Wang, Ching-wei
AU - Yurovskiy, Vladimir
AU - Maevskikh, Pavel
AU - Khanagha, Vahid
AU - Jiang, Yi
AU - Yu, Li
AU - Liu, Zhihong
AU - Li, Daiqiang
AU - Schuffler, Peter J.
AU - Yu, Qifeng
AU - Chen, Hui
AU - Tang, Yuling
AU - Litjens, Geert
T2 - IEEE Journal of Biomedical and Health Informatics
DA - 2021/02//
PY - 2021
DO - 10.1109/JBHI.2020.3039741
DP - DOI.org (Crossref)
VL - 25
IS - 2
SP - 429
EP - 440
J2 - IEEE J. Biomed. Health Inform.
SN - 2168-2194, 2168-2208
UR - https://ieeexplore.ieee.org/document/9265237/
Y2 - 2022/07/13/12:46:14
ER -

Flextilesource: An openseadragon extension for efficient whole-slide image visualization.

Peter J. Schüffler, Gamze Gokturk Ozcan, Hikmat Al-Ahmadie and Thomas J. Fuchs.

J Pathol Inform, vol. 12, 1, p. 31, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_flextilesource_2021,
    title = {Flextilesource: {An} openseadragon extension for efficient whole-slide image visualization},
    volume = {12},
    issn = {2153-3539},
    shorttitle = {Flextilesource},
    url = {https://doi.org/10.4103/jpi.jpi_13_21},
    doi = {10.4103/jpi.jpi_13_21},
    language = {en},
    number = {1},
    urldate = {2021-09-14},
    journal = {Journal of Pathology Informatics},
    author = {Sch\"uffler, Peter J. and Ozcan, Gamze Gokturk and Al-Ahmadie, Hikmat and Fuchs, Thomas J.},
    year = {2021},
    pages = {31},
}

Download Endnote/RIS citation

TY - JOUR
TI - Flextilesource: An openseadragon extension for efficient whole-slide image visualization
AU - Schüffler, Peter J.
AU - Ozcan, Gamze Gokturk
AU - Al-Ahmadie, Hikmat
AU - Fuchs, Thomas J.
T2 - Journal of Pathology Informatics
DA - 2021///
PY - 2021
DO - 10.4103/jpi.jpi_13_21
VL - 12
IS - 1
SP - 31
J2 - J Pathol Inform
LA - en
SN - 2153-3539
ST - Flextilesource
UR - https://doi.org/10.4103/jpi.jpi_13_21
Y2 - 2021/09/14/18:42:14
ER -

Beyond Classification: Whole Slide Tissue Histopathology Analysis By End-To-End Part Learning.

Chensu Xie, Hassan Muhammad, Chad M. Vanderbilt, Raul Caso, Dig Vijay Kumar Yarlagadda, Gabriele Campanella and Thomas J. Fuchs.

Proceedings of the Third Conference on Medical Imaging with Deep Learning, Medical Imaging with Deep Learning, vol. 121, p. 843-856, PMLR, 6-24-2020

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{xie_beyond_2020,
    address = {Montr\'eal},
    title = {Beyond {Classification}: {Whole} {Slide} {Tissue} {Histopathology} {Analysis} {By} {End}-{To}-{End} {Part} {Learning}},
    volume = {121},
    shorttitle = {{EPL}},
    url = {http://proceedings.mlr.press/v121/xie20a.html},
    abstract = {An emerging technology in cancer care and research is the use of histopathology whole slide images (WSI). Leveraging computation methods to aid in WSI assessment poses unique challenges. WSIs, being extremely high resolution giga-pixel images, cannot be directly processed by convolutional neural networks (CNN) due to huge computational cost. For this reason, state-of-the-art methods for WSI analysis adopt a two-stage approach where the training of a tile encoder is decoupled from the tile aggregation. This results in a trade-off between learning diverse and discriminative features. In contrast, we propose end-to-end part learning (EPL) which is able to learn diverse features while ensuring that learned features are discriminative. Each WSI is modeled as consisting of k groups of tiles with similar features, defined as parts. A loss with respect to the slide label is backpropagated through an integrated CNN model to k input tiles that are used to represent each part. Our experiments show that EPL is capable of clinical grade prediction of prostate and basal cell carcinoma. Further, we show that diverse discriminative features produced by EPL succeeds in multi-label classification of lung cancer architectural subtypes. Beyond classification, our method provides rich information of slides for high quality clinical decision support.},
    booktitle = {Proceedings of the {Third} {Conference} on {Medical} {Imaging} with {Deep} {Learning}},
    publisher = {PMLR},
    author = {Xie, Chensu and Muhammad, Hassan and Vanderbilt, Chad M. and Caso, Raul and Yarlagadda, Dig Vijay Kumar and Campanella, Gabriele and Fuchs, Thomas J.},
    month = jun,
    year = {2020},
    pages = {843--856},
}

Download Endnote/RIS citation

TY - CONF
TI - Beyond Classification: Whole Slide Tissue Histopathology Analysis By End-To-End Part Learning
AU - Xie, Chensu
AU - Muhammad, Hassan
AU - Vanderbilt, Chad M.
AU - Caso, Raul
AU - Yarlagadda, Dig Vijay Kumar
AU - Campanella, Gabriele
AU - Fuchs, Thomas J.
T2 - Medical Imaging with Deep Learning
AB - An emerging technology in cancer care and research is the use of histopathology whole slide images (WSI). Leveraging computation methods to aid in WSI assessment poses unique challenges. WSIs, being extremely high resolution giga-pixel images, cannot be directly processed by convolutional neural networks (CNN) due to huge computational cost. For this reason, state-of-the-art methods for WSI analysis adopt a two-stage approach where the training of a tile encoder is decoupled from the tile aggregation. This results in a trade-off between learning diverse and discriminative features. In contrast, we propose end-to-end part learning (EPL) which is able to learn diverse features while ensuring that learned features are discriminative. Each WSI is modeled as consisting of k groups of tiles with similar features, defined as parts. A loss with respect to the slide label is backpropagated through an integrated CNN model to k input tiles that are used to represent each part. Our experiments show that EPL is capable of clinical grade prediction of prostate and basal cell carcinoma. Further, we show that diverse discriminative features produced by EPL succeeds in multi-label classification of lung cancer architectural subtypes. Beyond classification, our method provides rich information of slides for high quality clinical decision support.
C1 - Montréal
C3 - Proceedings of the Third Conference on Medical Imaging with Deep Learning
DA - 2020/06/24/
PY - 2020
VL - 121
SP - 843
EP - 856
PB - PMLR
ST - EPL
UR - http://proceedings.mlr.press/v121/xie20a.html
ER -

An emerging technology in cancer care and research is the use of histopathology whole slide images (WSI). Leveraging computation methods to aid in WSI assessment poses unique challenges. WSIs, being extremely high resolution giga-pixel images, cannot be directly processed by convolutional neural networks (CNN) due to huge computational cost. For this reason, state-of-the-art methods for WSI analysis adopt a two-stage approach where the training of a tile encoder is decoupled from the tile aggregation. This results in a trade-off between learning diverse and discriminative features. In contrast, we propose end-to-end part learning (EPL) which is able to learn diverse features while ensuring that learned features are discriminative. Each WSI is modeled as consisting of k groups of tiles with similar features, defined as parts. A loss with respect to the slide label is backpropagated through an integrated CNN model to k input tiles that are used to represent each part. Our experiments show that EPL is capable of clinical grade prediction of prostate and basal cell carcinoma. Further, we show that diverse discriminative features produced by EPL succeeds in multi-label classification of lung cancer architectural subtypes. Beyond classification, our method provides rich information of slides for high quality clinical decision support.

Validation of a digital pathology system including remote review during the COVID-19 pandemic.

Matthew G. Hanna, Victor E. Reuter, Orly Ardon, David Kim, Sahussapont Joseph Sirintrapun, Peter J. Schüffler, Klaus J. Busam, Jennifer L. Sauter, Edi Brogi, Lee K. Tan, Bin Xu, Tejus Bale, Narasimhan P. Agaram, Laura H. Tang, Lora H. Ellenson, John Philip, Lorraine Corsale, Evangelos Stamelos, Maria A. Friedlander, Peter Ntiamoah, Marc Labasin, Christine England, David S. Klimstra and Meera Hameed.

Modern Pathology, vol. 33, p. 2115–2127, 2020-06-22

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{hanna_validation_2020,
    title = {Validation of a digital pathology system including remote review during the {COVID}-19 pandemic},
    volume = {33},
    copyright = {2020 The Author(s), under exclusive licence to United States \& Canadian Academy of Pathology},
    issn = {1530-0285},
    url = {https://www.nature.com/articles/s41379-020-0601-5},
    doi = {10.1038/s41379-020-0601-5},
    abstract = {Remote digital pathology allows healthcare systems to maintain pathology operations during public health emergencies. Existing Clinical Laboratory Improvement Amendments regulations require pathologists to electronically verify patient reports from a certified facility. During the 2019 pandemic of COVID-19 disease, caused by the SAR-CoV-2 virus, this requirement potentially exposes pathologists, their colleagues, and household members to the risk of becoming infected. Relaxation of government enforcement of this regulation allows pathologists to review and report pathology specimens from a remote, non-CLIA certified facility. The availability of digital pathology systems can facilitate remote microscopic diagnosis, although formal comprehensive (case-based) validation of remote digital diagnosis has not been reported. All glass slides representing routine clinical signout workload in surgical pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on an Aperio GT450 at ×40 equivalent resolution (0.26 µm/pixel). Twelve pathologists from nine surgical pathology subspecialties remotely reviewed and reported complete pathology cases using a digital pathology system from a non-CLIA certified facility through a secure connection. Whole slide images were integrated to and launched within the laboratory information system to a custom vendor-agnostic, whole slide image viewer. Remote signouts utilized consumer-grade computers and monitors (monitor size, 13.3–42 in.; resolution, 1280 × 800–3840 × 2160 pixels) connecting to an institution clinical workstation via secure virtual private network. Pathologists subsequently reviewed all corresponding glass slides using a light microscope within the CLIA-certified department. Intraobserver concordance metrics included reporting elements of top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and ancillary testing. The median whole slide image file size was 1.3 GB; scan time/slide averaged 90 s; and scanned tissue area averaged 612 mm2. Signout sessions included a total of 108 cases, comprised of 254 individual parts and 1196 slides. Major diagnostic equivalency was 100\% between digital and glass slide diagnoses; and overall concordance was 98.8\% (251/254). This study reports validation of primary diagnostic review and reporting of complete pathology cases from a remote site during a public health emergency. Our experience shows high (100\%) intraobserver digital to glass slide major diagnostic concordance when reporting from a remote site. This randomized, prospective study successfully validated remote use of a digital pathology system including operational feasibility supporting remote review and reporting of pathology specimens, and evaluation of remote access performance and usability for remote signout.},
    language = {en},
    urldate = {2020-06-22},
    journal = {Modern Pathology},
    author = {Hanna, Matthew G. and Reuter, Victor E. and Ardon, Orly and Kim, David and Sirintrapun, Sahussapont Joseph and Sch\"uffler, Peter J. and Busam, Klaus J. and Sauter, Jennifer L. and Brogi, Edi and Tan, Lee K. and Xu, Bin and Bale, Tejus and Agaram, Narasimhan P. and Tang, Laura H. and Ellenson, Lora H. and Philip, John and Corsale, Lorraine and Stamelos, Evangelos and Friedlander, Maria A. and Ntiamoah, Peter and Labasin, Marc and England, Christine and Klimstra, David S. and Hameed, Meera},
    month = jun,
    year = {2020},
    pages = {2115--2127},
}

Download Endnote/RIS citation

TY - JOUR
TI - Validation of a digital pathology system including remote review during the COVID-19 pandemic
AU - Hanna, Matthew G.
AU - Reuter, Victor E.
AU - Ardon, Orly
AU - Kim, David
AU - Sirintrapun, Sahussapont Joseph
AU - Schüffler, Peter J.
AU - Busam, Klaus J.
AU - Sauter, Jennifer L.
AU - Brogi, Edi
AU - Tan, Lee K.
AU - Xu, Bin
AU - Bale, Tejus
AU - Agaram, Narasimhan P.
AU - Tang, Laura H.
AU - Ellenson, Lora H.
AU - Philip, John
AU - Corsale, Lorraine
AU - Stamelos, Evangelos
AU - Friedlander, Maria A.
AU - Ntiamoah, Peter
AU - Labasin, Marc
AU - England, Christine
AU - Klimstra, David S.
AU - Hameed, Meera
T2 - Modern Pathology
AB - Remote digital pathology allows healthcare systems to maintain pathology operations during public health emergencies. Existing Clinical Laboratory Improvement Amendments regulations require pathologists to electronically verify patient reports from a certified facility. During the 2019 pandemic of COVID-19 disease, caused by the SAR-CoV-2 virus, this requirement potentially exposes pathologists, their colleagues, and household members to the risk of becoming infected. Relaxation of government enforcement of this regulation allows pathologists to review and report pathology specimens from a remote, non-CLIA certified facility. The availability of digital pathology systems can facilitate remote microscopic diagnosis, although formal comprehensive (case-based) validation of remote digital diagnosis has not been reported. All glass slides representing routine clinical signout workload in surgical pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on an Aperio GT450 at ×40 equivalent resolution (0.26 µm/pixel). Twelve pathologists from nine surgical pathology subspecialties remotely reviewed and reported complete pathology cases using a digital pathology system from a non-CLIA certified facility through a secure connection. Whole slide images were integrated to and launched within the laboratory information system to a custom vendor-agnostic, whole slide image viewer. Remote signouts utilized consumer-grade computers and monitors (monitor size, 13.3–42 in.; resolution, 1280 × 800–3840 × 2160 pixels) connecting to an institution clinical workstation via secure virtual private network. Pathologists subsequently reviewed all corresponding glass slides using a light microscope within the CLIA-certified department. Intraobserver concordance metrics included reporting elements of top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and ancillary testing. The median whole slide image file size was 1.3 GB; scan time/slide averaged 90 s; and scanned tissue area averaged 612 mm2. Signout sessions included a total of 108 cases, comprised of 254 individual parts and 1196 slides. Major diagnostic equivalency was 100% between digital and glass slide diagnoses; and overall concordance was 98.8% (251/254). This study reports validation of primary diagnostic review and reporting of complete pathology cases from a remote site during a public health emergency. Our experience shows high (100%) intraobserver digital to glass slide major diagnostic concordance when reporting from a remote site. This randomized, prospective study successfully validated remote use of a digital pathology system including operational feasibility supporting remote review and reporting of pathology specimens, and evaluation of remote access performance and usability for remote signout.
DA - 2020/06/22/
PY - 2020
DO - 10.1038/s41379-020-0601-5
DP - www.nature.com
VL - 33
SP - 2115
EP - 2127
LA - en
SN - 1530-0285
UR - https://www.nature.com/articles/s41379-020-0601-5
Y2 - 2020/06/22/12:46:57
ER -

Remote digital pathology allows healthcare systems to maintain pathology operations during public health emergencies. Existing Clinical Laboratory Improvement Amendments regulations require pathologists to electronically verify patient reports from a certified facility. During the 2019 pandemic of COVID-19 disease, caused by the SAR-CoV-2 virus, this requirement potentially exposes pathologists, their colleagues, and household members to the risk of becoming infected. Relaxation of government enforcement of this regulation allows pathologists to review and report pathology specimens from a remote, non-CLIA certified facility. The availability of digital pathology systems can facilitate remote microscopic diagnosis, although formal comprehensive (case-based) validation of remote digital diagnosis has not been reported. All glass slides representing routine clinical signout workload in surgical pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on an Aperio GT450 at ×40 equivalent resolution (0.26 µm/pixel). Twelve pathologists from nine surgical pathology subspecialties remotely reviewed and reported complete pathology cases using a digital pathology system from a non-CLIA certified facility through a secure connection. Whole slide images were integrated to and launched within the laboratory information system to a custom vendor-agnostic, whole slide image viewer. Remote signouts utilized consumer-grade computers and monitors (monitor size, 13.3–42 in.; resolution, 1280 × 800–3840 × 2160 pixels) connecting to an institution clinical workstation via secure virtual private network. Pathologists subsequently reviewed all corresponding glass slides using a light microscope within the CLIA-certified department. Intraobserver concordance metrics included reporting elements of top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and ancillary testing. The median whole slide image file size was 1.3 GB; scan time/slide averaged 90 s; and scanned tissue area averaged 612 mm2. Signout sessions included a total of 108 cases, comprised of 254 individual parts and 1196 slides. Major diagnostic equivalency was 100% between digital and glass slide diagnoses; and overall concordance was 98.8% (251/254). This study reports validation of primary diagnostic review and reporting of complete pathology cases from a remote site during a public health emergency. Our experience shows high (100%) intraobserver digital to glass slide major diagnostic concordance when reporting from a remote site. This randomized, prospective study successfully validated remote use of a digital pathology system including operational feasibility supporting remote review and reporting of pathology specimens, and evaluation of remote access performance and usability for remote signout.

(Re) Defining the high-power field for digital pathology.

David Kim, Liron Pantanowitz, Peter Schüffler, Dig Vijay Kumar Yarlagadda, Orly Ardon, Victor E. Reuter, Meera Hameed, David S. Klimstra and Matthew G. Hanna.

Journal of Pathology Informatics, vol. 11, 1, p. 33, 1/1/2020

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{kim_re_2020,
    title = {({Re}) {Defining} the high-power field for digital pathology},
    volume = {11},
    issn = {2153-3539},
    url = {https://www.jpathinformatics.org/article.asp?issn=2153-3539;year=2020;volume=11;issue=1;spage=33;epage=33;aulast=Kim;type=0},
    doi = {10.4103/jpi.jpi_48_20},
    abstract = {{\textless}br{\textgreater}\textbf{Background:} The microscope high-power field (HPF) is the cornerstone for histopathology diagnostic evaluation such as the quantification of mitotic figures, lymphocytes, and tumor grading. With traditional light microscopy, HPFs are typically evaluated by quantifying histologic events in 10 fields of view at × 400 magnification. In the era of digital pathology, new variables are introduced that may affect HPF evaluation. The aim of this study was to determine the parameters that influence HPF in whole slide images (WSIs). \textbf{Materials and Methods:} Glass slides scanned on various devices (Leica's Aperio GT450, AT2, and ScanScope XT; Philips UltraFast Scanner; Hamamatsu's Nanozoomer 2.0HT; and 3DHistech's P1000) were compared to acquired digital slides reviewed on each vendor's respective WSI viewer software (e.g., Aperio ImageScope, ImageScope DX, Philips IMS, 3DHistech CaseViewer, and Hamamatsu NDP.view) and an in-house developed vendor-agnostic viewer. WSIs were reviewed at “×40” equivalent HPF on different sized monitors with varying display resolutions (1900 × 1080–4500 × 3000) and aspect ratios (e.g., Food and Drug Administration [FDA]-cleared 27” Philips PS27QHDCR, FDA-cleared 24” Dell MR2416, 24” Hewlett Packard Z24n G2, and 28” Microsoft Surface Studio). Digital and microscopic HPF areas were calculated and compared. \textbf{Results:} A significant variation of HPF area occurred between differing monitor size and display resolutions with minor differences between WSI viewers. No differences were identified by scanner or WSIs scanned at different resolutions (e.g., 0.5, 0.25, 0.24, and 0.12 μm/pixel). \textbf{Conclusion:} Glass slide HPF at × 400 magnification with conventional light microscopy was not equivalent to “×40” digital HPF areas. Digital HPF quantification may vary due to differences in the tissue area displayed by monitor sizes, display resolutions, and WSI viewers but not by scanner or scanning resolution. These findings will need to be further clinically validated with potentially new digital metrics for evaluation.{\textless}br{\textgreater}},
    language = {en},
    number = {1},
    urldate = {2020-10-28},
    journal = {Journal of Pathology Informatics},
    author = {Kim, David and Pantanowitz, Liron and Sch\"uffler, Peter and Yarlagadda, Dig Vijay Kumar and Ardon, Orly and Reuter, Victor E. and Hameed, Meera and Klimstra, David S. and Hanna, Matthew G.},
    month = jan,
    year = {2020},
    Distributor: Medknow Publications and Media Pvt. Ltd.
    Institution: Medknow Publications and Media Pvt. Ltd.
    Label: Medknow Publications and Media Pvt. Ltd.
    Publisher: Medknow Publications},
    pages = {33},
}

Download Endnote/RIS citation

TY - JOUR
TI - (Re) Defining the high-power field for digital pathology
AU - Kim, David
AU - Pantanowitz, Liron
AU - Schüffler, Peter
AU - Yarlagadda, Dig Vijay Kumar
AU - Ardon, Orly
AU - Reuter, Victor E.
AU - Hameed, Meera
AU - Klimstra, David S.
AU - Hanna, Matthew G.
T2 - Journal of Pathology Informatics
AB -
Background: The microscope high-power field (HPF) is the cornerstone for histopathology diagnostic evaluation such as the quantification of mitotic figures, lymphocytes, and tumor grading. With traditional light microscopy, HPFs are typically evaluated by quantifying histologic events in 10 fields of view at × 400 magnification. In the era of digital pathology, new variables are introduced that may affect HPF evaluation. The aim of this study was to determine the parameters that influence HPF in whole slide images (WSIs). Materials and Methods: Glass slides scanned on various devices (Leica's Aperio GT450, AT2, and ScanScope XT; Philips UltraFast Scanner; Hamamatsu's Nanozoomer 2.0HT; and 3DHistech's P1000) were compared to acquired digital slides reviewed on each vendor's respective WSI viewer software (e.g., Aperio ImageScope, ImageScope DX, Philips IMS, 3DHistech CaseViewer, and Hamamatsu NDP.view) and an in-house developed vendor-agnostic viewer. WSIs were reviewed at “×40” equivalent HPF on different sized monitors with varying display resolutions (1900 × 1080–4500 × 3000) and aspect ratios (e.g., Food and Drug Administration [FDA]-cleared 27” Philips PS27QHDCR, FDA-cleared 24” Dell MR2416, 24” Hewlett Packard Z24n G2, and 28” Microsoft Surface Studio). Digital and microscopic HPF areas were calculated and compared. Results: A significant variation of HPF area occurred between differing monitor size and display resolutions with minor differences between WSI viewers. No differences were identified by scanner or WSIs scanned at different resolutions (e.g., 0.5, 0.25, 0.24, and 0.12 μm/pixel). Conclusion: Glass slide HPF at × 400 magnification with conventional light microscopy was not equivalent to “×40” digital HPF areas. Digital HPF quantification may vary due to differences in the tissue area displayed by monitor sizes, display resolutions, and WSI viewers but not by scanner or scanning resolution. These findings will need to be further clinically validated with potentially new digital metrics for evaluation.

DA - 2020/01/01/
PY - 2020
DO - 10.4103/jpi.jpi_48_20
DP - www.jpathinformatics.org
VL - 11
IS - 1
SP - 33
LA - en
SN - 2153-3539
UR - https://www.jpathinformatics.org/article.asp?issn=2153-3539;year=2020;volume=11;issue=1;spage=33;epage=33;aulast=Kim;type=0
Y2 - 2020/10/28/14:22:22
ER -

Background: The microscope high-power field (HPF) is the cornerstone for histopathology diagnostic evaluation such as the quantification of mitotic figures, lymphocytes, and tumor grading. With traditional light microscopy, HPFs are typically evaluated by quantifying histologic events in 10 fields of view at × 400 magnification. In the era of digital pathology, new variables are introduced that may affect HPF evaluation. The aim of this study was to determine the parameters that influence HPF in whole slide images (WSIs). Materials and Methods: Glass slides scanned on various devices (Leica's Aperio GT450, AT2, and ScanScope XT; Philips UltraFast Scanner; Hamamatsu's Nanozoomer 2.0HT; and 3DHistech's P1000) were compared to acquired digital slides reviewed on each vendor's respective WSI viewer software (e.g., Aperio ImageScope, ImageScope DX, Philips IMS, 3DHistech CaseViewer, and Hamamatsu NDP.view) and an in-house developed vendor-agnostic viewer. WSIs were reviewed at “×40” equivalent HPF on different sized monitors with varying display resolutions (1900 × 1080–4500 × 3000) and aspect ratios (e.g., Food and Drug Administration [FDA]-cleared 27” Philips PS27QHDCR, FDA-cleared 24” Dell MR2416, 24” Hewlett Packard Z24n G2, and 28” Microsoft Surface Studio). Digital and microscopic HPF areas were calculated and compared. Results: A significant variation of HPF area occurred between differing monitor size and display resolutions with minor differences between WSI viewers. No differences were identified by scanner or WSIs scanned at different resolutions (e.g., 0.5, 0.25, 0.24, and 0.12 μm/pixel). Conclusion: Glass slide HPF at × 400 magnification with conventional light microscopy was not equivalent to “×40” digital HPF areas. Digital HPF quantification may vary due to differences in the tissue area displayed by monitor sizes, display resolutions, and WSI viewers but not by scanner or scanning resolution. These findings will need to be further clinically validated with potentially new digital metrics for evaluation.

Deep Interactive Learning: An Efficient Labeling Approach for Deep Learning-Based Osteosarcoma Treatment Response Assessment.

David Joon Ho, Narasimhan P. Agaram, Peter J. Schüffler, Chad M. Vanderbilt, Marc-Henri Jean, Meera R. Hameed and Thomas J. Fuchs.

In: Anne L. Martel, Purang Abolmaesumi, Danail Stoyanov, Diana Mateus, Maria A. Zuluaga, S. Kevin Zhou, Daniel Racoceanu and Leo Joskowicz (eds.) Medical Image Computing and Computer Assisted Intervention – MICCAI 2020, vol. 12265, p. 540-549, Springer International Publishing, ISBN 978-3-030-59721-4 978-3-030-59722-1, 2020

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{martel_deep_2020,
    address = {Cham},
    title = {Deep {Interactive} {Learning}: {An} {Efficient} {Labeling} {Approach} for {Deep} {Learning}-{Based} {Osteosarcoma} {Treatment} {Response} {Assessment}},
    volume = {12265},
    isbn = {978-3-030-59721-4 978-3-030-59722-1},
    shorttitle = {Deep {Interactive} {Learning}},
    url = {http://link.springer.com/10.1007/978-3-030-59722-1_52},
    language = {en},
    urldate = {2020-10-06},
    booktitle = {Medical {Image} {Computing} and {Computer} {Assisted} {Intervention} – {MICCAI} 2020},
    publisher = {Springer International Publishing},
    author = {Ho, David Joon and Agaram, Narasimhan P. and Sch\"uffler, Peter J. and Vanderbilt, Chad M. and Jean, Marc-Henri and Hameed, Meera R. and Fuchs, Thomas J.},
    editor = {Martel, Anne L. and Abolmaesumi, Purang and Stoyanov, Danail and Mateus, Diana and Zuluaga, Maria A. and Zhou, S. Kevin and Racoceanu, Daniel and Joskowicz, Leo},
    year = {2020},
    doi = {10.1007/978-3-030-59722-1_52},
    pages = {540--549},
}

Download Endnote/RIS citation

TY - CHAP
TI - Deep Interactive Learning: An Efficient Labeling Approach for Deep Learning-Based Osteosarcoma Treatment Response Assessment
AU - Ho, David Joon
AU - Agaram, Narasimhan P.
AU - Schüffler, Peter J.
AU - Vanderbilt, Chad M.
AU - Jean, Marc-Henri
AU - Hameed, Meera R.
AU - Fuchs, Thomas J.
T2 - Medical Image Computing and Computer Assisted Intervention – MICCAI 2020
A2 - Martel, Anne L.
A2 - Abolmaesumi, Purang
A2 - Stoyanov, Danail
A2 - Mateus, Diana
A2 - Zuluaga, Maria A.
A2 - Zhou, S. Kevin
A2 - Racoceanu, Daniel
A2 - Joskowicz, Leo
CY - Cham
DA - 2020///
PY - 2020
DP - DOI.org (Crossref)
VL - 12265
SP - 540
EP - 549
LA - en
PB - Springer International Publishing
SN - 978-3-030-59721-4 978-3-030-59722-1
ST - Deep Interactive Learning
UR - http://link.springer.com/10.1007/978-3-030-59722-1_52
Y2 - 2020/10/06/08:47:12
ER -

Unsupervised Subtyping of Cholangiocarcinoma Using a Deep Clustering Convolutional Autoencoder.

Hassan Muhammad, Carlie S. Sigel, Gabriele Campanella, Thomas Boerner, Linda M. Pak, Stefan Büttner, Jan N. M. IJzermans, Bas Groot Koerkamp, Michael Doukas, William R. Jarnagin, Amber L. Simpson and Thomas J. Fuchs.

In: Dinggang Shen, Tianming Liu, Terry M. Peters, Lawrence H. Staib, Caroline Essert, Sean Zhou, Pew-Thian Yap and Ali Khan (eds.) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019, vol. 11764, p. 604-612, Springer International Publishing, ISBN 978-3-030-32238-0 978-3-030-32239-7, 2019-10-30

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{shen_unsupervised_2019,
    address = {Cham},
    title = {Unsupervised {Subtyping} of {Cholangiocarcinoma} {Using} a {Deep} {Clustering} {Convolutional} {Autoencoder}},
    volume = {11764},
    isbn = {978-3-030-32238-0 978-3-030-32239-7},
    url = {http://link.springer.com/10.1007/978-3-030-32239-7_67},
    abstract = {Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is diﬃcult and largely undeﬁned because of small sample sizes, the sheer volume of time and experience needed to undertake such a task, and the inherent diﬃculty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 digitized whole slides, based on visual similarity. Clusters based on this visual dictionary of histologic patterns are interpreted as new ICC subtypes and evaluated by training Cox-proportional hazard survival models, resulting in statistically signiﬁcant patient stratiﬁcation.},
    language = {en},
    urldate = {2019-11-20},
    booktitle = {Medical {Image} {Computing} and {Computer} {Assisted} {Intervention} – {MICCAI} 2019},
    publisher = {Springer International Publishing},
    author = {Muhammad, Hassan and Sigel, Carlie S. and Campanella, Gabriele and Boerner, Thomas and Pak, Linda M. and B\"uttner, Stefan and IJzermans, Jan N. M. and Koerkamp, Bas Groot and Doukas, Michael and Jarnagin, William R. and Simpson, Amber L. and Fuchs, Thomas J.},
    editor = {Shen, Dinggang and Liu, Tianming and Peters, Terry M. and Staib, Lawrence H. and Essert, Caroline and Zhou, Sean and Yap, Pew-Thian and Khan, Ali},
    month = oct,
    year = {2019},
    doi = {10.1007/978-3-030-32239-7_67},
    pages = {604--612},
}

Download Endnote/RIS citation

TY - CHAP
TI - Unsupervised Subtyping of Cholangiocarcinoma Using a Deep Clustering Convolutional Autoencoder
AU - Muhammad, Hassan
AU - Sigel, Carlie S.
AU - Campanella, Gabriele
AU - Boerner, Thomas
AU - Pak, Linda M.
AU - Büttner, Stefan
AU - IJzermans, Jan N. M.
AU - Koerkamp, Bas Groot
AU - Doukas, Michael
AU - Jarnagin, William R.
AU - Simpson, Amber L.
AU - Fuchs, Thomas J.
T2 - Medical Image Computing and Computer Assisted Intervention – MICCAI 2019
A2 - Shen, Dinggang
A2 - Liu, Tianming
A2 - Peters, Terry M.
A2 - Staib, Lawrence H.
A2 - Essert, Caroline
A2 - Zhou, Sean
A2 - Yap, Pew-Thian
A2 - Khan, Ali
AB - Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is diﬃcult and largely undeﬁned because of small sample sizes, the sheer volume of time and experience needed to undertake such a task, and the inherent diﬃculty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 digitized whole slides, based on visual similarity. Clusters based on this visual dictionary of histologic patterns are interpreted as new ICC subtypes and evaluated by training Cox-proportional hazard survival models, resulting in statistically signiﬁcant patient stratiﬁcation.
CY - Cham
DA - 2019/10/30/
PY - 2019
DP - Crossref
VL - 11764
SP - 604
EP - 612
LA - en
PB - Springer International Publishing
SN - 978-3-030-32238-0 978-3-030-32239-7
UR - http://link.springer.com/10.1007/978-3-030-32239-7_67
Y2 - 2019/11/20/20:24:50
ER -

Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is diﬃcult and largely undeﬁned because of small sample sizes, the sheer volume of time and experience needed to undertake such a task, and the inherent diﬃculty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 digitized whole slides, based on visual similarity. Clusters based on this visual dictionary of histologic patterns are interpreted as new ICC subtypes and evaluated by training Cox-proportional hazard survival models, resulting in statistically signiﬁcant patient stratiﬁcation.

Accuracy of Intraoperative Frozen Section of Sentinel Lymph Nodes After Neoadjuvant Chemotherapy for Breast Carcinoma.

Anne Grabenstetter, Tracy-Ann Moo, Sabina Hajiyeva, Peter J. Schüffler, Pallavi Khattar, Maria A. Friedlander, Maura A. McCormack, Monica Raiss, Emily C. Zabor, Andrea Barrio, Monica Morrow and Marcia Edelweiss.

Am. J. Surg. Pathol., Jun 18, 2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{grabenstetter_accuracy_2019,
    title = {Accuracy of {Intraoperative} {Frozen} {Section} of {Sentinel} {Lymph} {Nodes} {After} {Neoadjuvant} {Chemotherapy} for {Breast} {Carcinoma}},
    issn = {1532-0979},
    url = {https://pubmed.ncbi.nlm.nih.gov/31219817/},
    doi = {10.1097/PAS.0000000000001311},
    abstract = {False-negative (FN) intraoperative frozen section (FS) results of sentinel lymph nodes (SLN) have been reported to be more common after neoadjuvant chemotherapy (NAC) in the primary surgical setting. We evaluated SLN FS assessment in breast cancer patients treated with NAC to determine the FN rate and the histomorphologic factors associated with FN results. Patients who had FS SLN assessment following NAC from July 2008 to July 2017 were identified. Of the 711 SLN FS cases, 522 were negative, 181 positive, and 8 deferred. The FN rate was 5.4\% (28/522). There were no false-positive results. Of the 8 deferred cases, 5 were positive on permanent section and 3 were negative. There was a higher frequency of micrometastasis and isolated tumor cells in FN cases (P{\textless}0.001). There was a significant increase in tissue surface area present on permanent section slides compared with FS slides (P{\textless}0.001), highlighting the inherent technical limitations of FS and histologic under-sampling of tissue which leads to most FN results. The majority (25/28, 89\%) of FN cases had metastatic foci identified exclusively on permanent sections and were not due to a true diagnostic interpretation error. FN cases were more frequently estrogen receptor positive (P{\textless}0.001), progesterone receptor positive (P=0.001), human epidermal growth factor receptor-2 negative (P=0.009) and histologic grade 1 (P=0.015), which most likely reflects the lower rates of pathologic complete response in these tumors. Despite its limitations, FS is a reliable modality to assess the presence of SLN metastases in NAC treated patients.},
    language = {eng},
    journal = {The American Journal of Surgical Pathology},
    author = {Grabenstetter, Anne and Moo, Tracy-Ann and Hajiyeva, Sabina and Sch\"uffler, Peter J. and Khattar, Pallavi and Friedlander, Maria A. and McCormack, Maura A. and Raiss, Monica and Zabor, Emily C. and Barrio, Andrea and Morrow, Monica and Edelweiss, Marcia},
    month = jun,
    year = {2019},
    pmid = {31219817},
}

Download Endnote/RIS citation

TY - JOUR
TI - Accuracy of Intraoperative Frozen Section of Sentinel Lymph Nodes After Neoadjuvant Chemotherapy for Breast Carcinoma
AU - Grabenstetter, Anne
AU - Moo, Tracy-Ann
AU - Hajiyeva, Sabina
AU - Schüffler, Peter J.
AU - Khattar, Pallavi
AU - Friedlander, Maria A.
AU - McCormack, Maura A.
AU - Raiss, Monica
AU - Zabor, Emily C.
AU - Barrio, Andrea
AU - Morrow, Monica
AU - Edelweiss, Marcia
T2 - The American Journal of Surgical Pathology
AB - False-negative (FN) intraoperative frozen section (FS) results of sentinel lymph nodes (SLN) have been reported to be more common after neoadjuvant chemotherapy (NAC) in the primary surgical setting. We evaluated SLN FS assessment in breast cancer patients treated with NAC to determine the FN rate and the histomorphologic factors associated with FN results. Patients who had FS SLN assessment following NAC from July 2008 to July 2017 were identified. Of the 711 SLN FS cases, 522 were negative, 181 positive, and 8 deferred. The FN rate was 5.4% (28/522). There were no false-positive results. Of the 8 deferred cases, 5 were positive on permanent section and 3 were negative. There was a higher frequency of micrometastasis and isolated tumor cells in FN cases (P<0.001). There was a significant increase in tissue surface area present on permanent section slides compared with FS slides (P<0.001), highlighting the inherent technical limitations of FS and histologic under-sampling of tissue which leads to most FN results. The majority (25/28, 89%) of FN cases had metastatic foci identified exclusively on permanent sections and were not due to a true diagnostic interpretation error. FN cases were more frequently estrogen receptor positive (P<0.001), progesterone receptor positive (P=0.001), human epidermal growth factor receptor-2 negative (P=0.009) and histologic grade 1 (P=0.015), which most likely reflects the lower rates of pathologic complete response in these tumors. Despite its limitations, FS is a reliable modality to assess the presence of SLN metastases in NAC treated patients.
DA - 2019/06/18/
PY - 2019
DO - 10.1097/PAS.0000000000001311
DP - PubMed
J2 - Am. J. Surg. Pathol.
LA - eng
SN - 1532-0979
UR - https://pubmed.ncbi.nlm.nih.gov/31219817/
ER -

False-negative (FN) intraoperative frozen section (FS) results of sentinel lymph nodes (SLN) have been reported to be more common after neoadjuvant chemotherapy (NAC) in the primary surgical setting. We evaluated SLN FS assessment in breast cancer patients treated with NAC to determine the FN rate and the histomorphologic factors associated with FN results. Patients who had FS SLN assessment following NAC from July 2008 to July 2017 were identified. Of the 711 SLN FS cases, 522 were negative, 181 positive, and 8 deferred. The FN rate was 5.4% (28/522). There were no false-positive results. Of the 8 deferred cases, 5 were positive on permanent section and 3 were negative. There was a higher frequency of micrometastasis and isolated tumor cells in FN cases (P<0.001). There was a significant increase in tissue surface area present on permanent section slides compared with FS slides (P<0.001), highlighting the inherent technical limitations of FS and histologic under-sampling of tissue which leads to most FN results. The majority (25/28, 89%) of FN cases had metastatic foci identified exclusively on permanent sections and were not due to a true diagnostic interpretation error. FN cases were more frequently estrogen receptor positive (P<0.001), progesterone receptor positive (P=0.001), human epidermal growth factor receptor-2 negative (P=0.009) and histologic grade 1 (P=0.015), which most likely reflects the lower rates of pathologic complete response in these tumors. Despite its limitations, FS is a reliable modality to assess the presence of SLN metastases in NAC treated patients.

VOCA: Cell Nuclei Detection In Histopathology Images By Vector Oriented Confidence Accumulation.

Chensu Xie, Chad M. Vanderbilt, Anne Grabenstetter and Thomas J. Fuchs.

Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, MIDL, vol. 102, p. 527-539, PMLR, 5/23/2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{xie_voca:_2019,
    address = {London},
    title = {{VOCA}: {Cell} {Nuclei} {Detection} {In} {Histopathology} {Images} {By} {Vector} {Oriented} {Confidence} {Accumulation}},
    volume = {102},
    shorttitle = {{VOCA}},
    url = {http://proceedings.mlr.press/v102/xie19a.html},
    abstract = {Cell nuclei detection is the basis for many tasks in Computational Pathology ranging from cancer diagnosis to survival analysis. It is a challenging task due to the significant inter/intra-class variation of cellular morphology. The problem is aggravated by the need for additional accurate localization of the nuclei for downstream applications. Most of the existing methods regress the probability of each pixel being a nuclei centroid, while relying on post-processing to implicitly infer the rough location of nuclei centers. To solve this problem we propose a novel multi-task learning framework called vector oriented confidence accumulation (VOCA) based on deep convolutional encoder-decoder. The model learns a confidence score, localization vector and weight of contribution for each pixel. The three tasks are trained concurrently and the confidence of pixels are accumulated according to the localization vectors in detection stage to generate a sparse map that describes accurate and precise cell locations. A detailed comparison to the state-of-the-art based on a publicly available colorectal cancer dataset showed superior detection performance and significantly higher localization accuracy.},
    booktitle = {Proceedings of {The} 2nd {International} {Conference} on {Medical} {Imaging} with {Deep} {Learning}},
    publisher = {PMLR},
    author = {Xie, Chensu and Vanderbilt, Chad M. and Grabenstetter, Anne and Fuchs, Thomas J.},
    month = may,
    year = {2019},
    pages = {527--539},
}

Download Endnote/RIS citation

TY - CONF
TI - VOCA: Cell Nuclei Detection In Histopathology Images By Vector Oriented Confidence Accumulation
AU - Xie, Chensu
AU - Vanderbilt, Chad M.
AU - Grabenstetter, Anne
AU - Fuchs, Thomas J.
T2 - MIDL
AB - Cell nuclei detection is the basis for many tasks in Computational Pathology ranging from cancer diagnosis to survival analysis. It is a challenging task due to the significant inter/intra-class variation of cellular morphology. The problem is aggravated by the need for additional accurate localization of the nuclei for downstream applications. Most of the existing methods regress the probability of each pixel being a nuclei centroid, while relying on post-processing to implicitly infer the rough location of nuclei centers. To solve this problem we propose a novel multi-task learning framework called vector oriented confidence accumulation (VOCA) based on deep convolutional encoder-decoder. The model learns a confidence score, localization vector and weight of contribution for each pixel. The three tasks are trained concurrently and the confidence of pixels are accumulated according to the localization vectors in detection stage to generate a sparse map that describes accurate and precise cell locations. A detailed comparison to the state-of-the-art based on a publicly available colorectal cancer dataset showed superior detection performance and significantly higher localization accuracy.
C1 - London
C3 - Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning
DA - 2019/05/23/
PY - 2019
VL - 102
SP - 527
EP - 539
PB - PMLR
ST - VOCA
UR - http://proceedings.mlr.press/v102/xie19a.html
ER -

Cell nuclei detection is the basis for many tasks in Computational Pathology ranging from cancer diagnosis to survival analysis. It is a challenging task due to the significant inter/intra-class variation of cellular morphology. The problem is aggravated by the need for additional accurate localization of the nuclei for downstream applications. Most of the existing methods regress the probability of each pixel being a nuclei centroid, while relying on post-processing to implicitly infer the rough location of nuclei centers. To solve this problem we propose a novel multi-task learning framework called vector oriented confidence accumulation (VOCA) based on deep convolutional encoder-decoder. The model learns a confidence score, localization vector and weight of contribution for each pixel. The three tasks are trained concurrently and the confidence of pixels are accumulated according to the localization vectors in detection stage to generate a sparse map that describes accurate and precise cell locations. A detailed comparison to the state-of-the-art based on a publicly available colorectal cancer dataset showed superior detection performance and significantly higher localization accuracy.

DeepPET: A deep encoder–decoder network for directly solving the PET image reconstruction inverse problem.

Ida Häggström, C. Ross Schmidtlein, Gabriele Campanella and Thomas J. Fuchs.

Medical Image Analysis, vol. 54, p. 253-262, 2019-03-30

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{haggstrom_deeppet:_2019,
    title = {{DeepPET}: {A} deep encoder–decoder network for directly solving the {PET} image
    reconstruction inverse problem},
    volume = {54},
    url = {https://doi.org/10.1016/j.media.2019.03.013},
    doi = {10.1016/j.media.2019.03.013},
    abstract = {The purpose of this research was to implement a deep learning network to overcome two of the major bottlenecks in improved image reconstruction for clinical positron emission tomography (PET). These are the lack of an automated means for the optimization of advanced image reconstruction algorithms, and the computational expense associated with these state-of-the art methods.
    We thus present a novel end-to-end PET image reconstruction technique, called DeepPET, based on a deep convolutional
    encoder–decoder network, which takes PET sinogram data as input and directly and quickly outputs high quality, quantitative PET images. Using simulated data derived from a whole-body digital phantom, we randomly sampled the configurable parameters to generate realistic images, which were each augmented to a total of more than 291,000 reference images. Realistic PET acquisitions
    of these images were simulated, resulting in noisy sinogram data, used for training, validation, and testing the DeepPET network.
    We demonstrated that DeepPET generates higher quality images compared to conventional techniques, in terms of relative root mean squared error (11\%/53\% lower than ordered subset expectation maximization (OSEM)/filtered back-projection (FBP), structural similarity index (1\%/11\% higher than OSEM/FBP), and peak signal-to-noise ratio (1.1/3.8 dB higher than OSEM/FBP). In addition, we show that DeepPET reconstructs images 108 and 3 times faster than OSEM and FBP, respectively. Finally, DeepPET
    was successfully applied to real clinical data. This study shows that an end-to-end encoder–decoder network can produce high quality PET images at a fraction of the time compared to conventional methods.},
    language = {en},
    journal = {Medical Image Analysis},
    author = {H\"aggstr\"om, Ida and Schmidtlein, C. Ross and Campanella, Gabriele and Fuchs, Thomas J.},
    month = mar,
    year = {2019},
    keywords = {Computer Vision and Pattern Recognition},
    pages = {253--262},
}

Download Endnote/RIS citation

TY - JOUR
TI - DeepPET: A deep encoder–decoder network for directly solving the PET image
reconstruction inverse problem
AU - Häggström, Ida
AU - Schmidtlein, C. Ross
AU - Campanella, Gabriele
AU - Fuchs, Thomas J.
T2 - Medical Image Analysis
AB - The purpose of this research was to implement a deep learning network to overcome two of the major bottlenecks in improved image reconstruction for clinical positron emission tomography (PET). These are the lack of an automated means for the optimization of advanced image reconstruction algorithms, and the computational expense associated with these state-of-the art methods.
We thus present a novel end-to-end PET image reconstruction technique, called DeepPET, based on a deep convolutional
encoder–decoder network, which takes PET sinogram data as input and directly and quickly outputs high quality, quantitative PET images. Using simulated data derived from a whole-body digital phantom, we randomly sampled the configurable parameters to generate realistic images, which were each augmented to a total of more than 291,000 reference images. Realistic PET acquisitions
of these images were simulated, resulting in noisy sinogram data, used for training, validation, and testing the DeepPET network.
We demonstrated that DeepPET generates higher quality images compared to conventional techniques, in terms of relative root mean squared error (11%/53% lower than ordered subset expectation maximization (OSEM)/filtered back-projection (FBP), structural similarity index (1%/11% higher than OSEM/FBP), and peak signal-to-noise ratio (1.1/3.8 dB higher than OSEM/FBP). In addition, we show that DeepPET reconstructs images 108 and 3 times faster than OSEM and FBP, respectively. Finally, DeepPET
was successfully applied to real clinical data. This study shows that an end-to-end encoder–decoder network can produce high quality PET images at a fraction of the time compared to conventional methods.
DA - 2019/03/30/
PY - 2019
DO - 10.1016/j.media.2019.03.013
VL - 54
SP - 253
EP - 262
LA - en
UR - https://doi.org/10.1016/j.media.2019.03.013
KW - Computer Vision and Pattern Recognition
ER -

The purpose of this research was to implement a deep learning network to overcome two of the major bottlenecks in improved image reconstruction for clinical positron emission tomography (PET). These are the lack of an automated means for the optimization of advanced image reconstruction algorithms, and the computational expense associated with these state-of-the art methods. We thus present a novel end-to-end PET image reconstruction technique, called DeepPET, based on a deep convolutional encoder–decoder network, which takes PET sinogram data as input and directly and quickly outputs high quality, quantitative PET images. Using simulated data derived from a whole-body digital phantom, we randomly sampled the configurable parameters to generate realistic images, which were each augmented to a total of more than 291,000 reference images. Realistic PET acquisitions of these images were simulated, resulting in noisy sinogram data, used for training, validation, and testing the DeepPET network. We demonstrated that DeepPET generates higher quality images compared to conventional techniques, in terms of relative root mean squared error (11%/53% lower than ordered subset expectation maximization (OSEM)/filtered back-projection (FBP), structural similarity index (1%/11% higher than OSEM/FBP), and peak signal-to-noise ratio (1.1/3.8 dB higher than OSEM/FBP). In addition, we show that DeepPET reconstructs images 108 and 3 times faster than OSEM and FBP, respectively. Finally, DeepPET was successfully applied to real clinical data. This study shows that an end-to-end encoder–decoder network can produce high quality PET images at a fraction of the time compared to conventional methods.

Towards Unsupervised Cancer Subtyping: Predicting Prognosis Using A Histologic Visual Dictionary.

arXiv:1903.05257 [cs, q-bio, stat], 2019-03-12

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{muhammad_towards_2019,
    title = {Towards {Unsupervised} {Cancer} {Subtyping}: {Predicting} {Prognosis} {Using} {A} {Histologic} {Visual} {Dictionary}},
    shorttitle = {Towards {Unsupervised} {Cancer} {Subtyping}},
    url = {http://arxiv.org/abs/1903.05257},
    abstract = {Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is difficult and largely undefined because of small sample sizes, the sheer volume of time needed to undertake on such a task, and the inherent difficulty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 ICC digitized whole slides, based on visual similarity. From this visual dictionary of histologic patterns, we use the clusters as covariates to train Cox-proportional hazard survival models. In univariate analysis, three clusters were significantly associated with recurrence-free survival. Combinations of these clusters were significant in multivariate analysis. In a multivariate analysis of all clusters, five showed significance to recurrence-free survival, however the overall model was not measured to be significant. Finally, a pathologist assigned clinical terminology to the significant clusters in the visual dictionary and found evidence supporting the hypothesis that collagen-enriched fibrosis plays a role in disease severity. These results offer insight into the future of cancer subtyping and show that computational pathology can contribute to disease prognostication, especially in rare cancers.},
    urldate = {2019-04-09},
    journal = {arXiv:1903.05257 [cs, q-bio, stat]},
    author = {Muhammad, Hassan and Sigel, Carlie S. and Campanella, Gabriele and Boerner, Thomas and Pak, Linda M. and B\"uttner, Stefan and IJzermans, Jan N. M. and Koerkamp, Bas Groot and Doukas, Michael and Jarnagin, William R. and Simpson, Amber and Fuchs, Thomas J.},
    month = mar,
    year = {2019},
    keywords = {Computer Science - Computer Vision and Pattern Recognition, Quantitative Biology - Tissues and Organs, Statistics - Machine Learning},
}

Download Endnote/RIS citation

TY - JOUR
TI - Towards Unsupervised Cancer Subtyping: Predicting Prognosis Using A Histologic Visual Dictionary
AU - Muhammad, Hassan
AU - Sigel, Carlie S.
AU - Campanella, Gabriele
AU - Boerner, Thomas
AU - Pak, Linda M.
AU - Büttner, Stefan
AU - IJzermans, Jan N. M.
AU - Koerkamp, Bas Groot
AU - Doukas, Michael
AU - Jarnagin, William R.
AU - Simpson, Amber
AU - Fuchs, Thomas J.
T2 - arXiv:1903.05257 [cs, q-bio, stat]
AB - Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is difficult and largely undefined because of small sample sizes, the sheer volume of time needed to undertake on such a task, and the inherent difficulty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 ICC digitized whole slides, based on visual similarity. From this visual dictionary of histologic patterns, we use the clusters as covariates to train Cox-proportional hazard survival models. In univariate analysis, three clusters were significantly associated with recurrence-free survival. Combinations of these clusters were significant in multivariate analysis. In a multivariate analysis of all clusters, five showed significance to recurrence-free survival, however the overall model was not measured to be significant. Finally, a pathologist assigned clinical terminology to the significant clusters in the visual dictionary and found evidence supporting the hypothesis that collagen-enriched fibrosis plays a role in disease severity. These results offer insight into the future of cancer subtyping and show that computational pathology can contribute to disease prognostication, especially in rare cancers.
DA - 2019/03/12/
PY - 2019
DP - arXiv.org
ST - Towards Unsupervised Cancer Subtyping
UR - http://arxiv.org/abs/1903.05257
Y2 - 2019/04/09/20:51:44
KW - Computer Science - Computer Vision and Pattern Recognition
KW - Quantitative Biology - Tissues and Organs
KW - Statistics - Machine Learning
ER -

Unlike common cancers, such as those of the prostate and breast, tumor grading in rare cancers is difficult and largely undefined because of small sample sizes, the sheer volume of time needed to undertake on such a task, and the inherent difficulty of extracting human-observed patterns. One of the most challenging examples is intrahepatic cholangiocarcinoma (ICC), a primary liver cancer arising from the biliary system, for which there is well-recognized tumor heterogeneity and no grading paradigm or prognostic biomarkers. In this paper, we propose a new unsupervised deep convolutional autoencoder-based clustering model that groups together cellular and structural morphologies of tumor in 246 ICC digitized whole slides, based on visual similarity. From this visual dictionary of histologic patterns, we use the clusters as covariates to train Cox-proportional hazard survival models. In univariate analysis, three clusters were significantly associated with recurrence-free survival. Combinations of these clusters were significant in multivariate analysis. In a multivariate analysis of all clusters, five showed significance to recurrence-free survival, however the overall model was not measured to be significant. Finally, a pathologist assigned clinical terminology to the significant clusters in the visual dictionary and found evidence supporting the hypothesis that collagen-enriched fibrosis plays a role in disease severity. These results offer insight into the future of cancer subtyping and show that computational pathology can contribute to disease prognostication, especially in rare cancers.

Whole slide imaging equivalency and efficiency study: experience at a large academic center.

Matthew G. Hanna, Victor E. Reuter, Meera R. Hameed, Lee K. Tan, Sarah Chiang, Carlie Sigel, Travis Hollmann, Dilip Giri, Jennifer Samboy, Carlos Moradel, Andrea Rosado, John R. Otilano, Christine England, Lorraine Corsale, Evangelos Stamelos, Yukako Yagi, Peter J. Schüffler, Thomas Fuchs, David S. Klimstra and S. Joseph Sirintrapun.

Modern Pathology, vol. 32, p. 916–928, 2019-02-18

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{hanna_whole_2019,
    title = {Whole slide imaging equivalency and efficiency study: experience at a large academic center},
    volume = {32},
    copyright = {2019 United States \& Canadian Academy of Pathology},
    issn = {1530-0285},
    shorttitle = {Whole slide imaging equivalency and efficiency study},
    url = {https://www.nature.com/articles/s41379-019-0205-0},
    doi = {10.1038/s41379-019-0205-0},
    abstract = {Whole slide imaging is Food and Drug Administration-approved for primary diagnosis in the United States of America; however, relatively few pathology departments in the country have fully implemented an enterprise wide digital pathology system enabled for primary diagnosis. Digital pathology has significant potential to transform pathology practice with several published studies documenting some level of diagnostic equivalence between digital and conventional systems. However, whole slide imaging also has significant potential to disrupt pathology practice, due to the differences in efficiency of manipulating digital images vis-\`a-vis glass slides, and studies on the efficiency of actual digital pathology workload are lacking. Our randomized, equivalency and efficiency study aimed to replicate clinical workflow, comparing conventional microscopy to a complete digital pathology signout using whole slide images, evaluating the equivalency and efficiency of glass slide to whole slide image reporting, reflective of true pathology practice workloads in the clinical setting. All glass slides representing an entire day’s routine clinical signout workload for six different anatomic pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on Leica Aperio AT2 at ×40 (0.25 µm/pixel). Integration of whole slide images for each accessioned case is through an interface between the Leica eSlide manager database and the laboratory information system, Cerner CoPathPlus. Pathologists utilized a standard institution computer workstation and viewed whole slide images through an internally developed, vendor agnostic whole slide image viewer, named the “MSK Slide Viewer”. Subspecialized pathologists first reported on glass slides from surgical pathology cases using routine clinical workflow. Glass slides were de-identified, scanned, and re-accessioned in the laboratory information system test environment. After a washout period of 13 weeks, pathologists reported the same clinical workload using whole slide image integrated within the laboratory information system. Intraobserver equivalency metrics included top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and the need to order ancillary testing (i.e., recuts, immunohistochemistry). Turnaround time (efficiency) evaluation was defined by the start of each case when opened in the laboratory information system and when the case was completed for that day (i.e., case sent to signout queue or pending ancillary studies). Eight pathologists participated from the following subspecialties: bone and soft tissue, genitourinary, gastrointestinal, breast, gynecologic, and dermatopathology. Glass slides signouts comprised of 204 cases, encompassing 2091 glass slides; and digital signouts comprised of 199 cases, encompassing 2073 whole slide images. The median whole slide image file size was 1.54 GB; scan time/slide, 6 min 24 s; and scan area 32.1 × 18.52 mm. Overall diagnostic equivalency (e.g., top-line diagnosis) was 99.3\% between digital and glass slide signout; however, signout using whole slide images showed a median overall 19\% decrease in efficiency per case. No significant difference by reader, subspecialty, or specimen type was identified. Our experience is the most comprehensive study to date and shows high intraobserver whole slide image to glass slide equivalence in reporting of true clinical workflows and workloads. Efficiency needs to improve for digital pathology to gain more traction among pathologists.},
    language = {En},
    urldate = {2019-02-21},
    journal = {Modern Pathology},
    author = {Hanna, Matthew G. and Reuter, Victor E. and Hameed, Meera R. and Tan, Lee K. and Chiang, Sarah and Sigel, Carlie and Hollmann, Travis and Giri, Dilip and Samboy, Jennifer and Moradel, Carlos and Rosado, Andrea and Otilano, John R. and England, Christine and Corsale, Lorraine and Stamelos, Evangelos and Yagi, Yukako and Sch\"uffler, Peter J. and Fuchs, Thomas and Klimstra, David S. and Sirintrapun, S. Joseph},
    month = feb,
    year = {2019},
    pages = {916--928},
}

Download Endnote/RIS citation

TY - JOUR
TI - Whole slide imaging equivalency and efficiency study: experience at a large academic center
AU - Hanna, Matthew G.
AU - Reuter, Victor E.
AU - Hameed, Meera R.
AU - Tan, Lee K.
AU - Chiang, Sarah
AU - Sigel, Carlie
AU - Hollmann, Travis
AU - Giri, Dilip
AU - Samboy, Jennifer
AU - Moradel, Carlos
AU - Rosado, Andrea
AU - Otilano, John R.
AU - England, Christine
AU - Corsale, Lorraine
AU - Stamelos, Evangelos
AU - Yagi, Yukako
AU - Schüffler, Peter J.
AU - Fuchs, Thomas
AU - Klimstra, David S.
AU - Sirintrapun, S. Joseph
T2 - Modern Pathology
AB - Whole slide imaging is Food and Drug Administration-approved for primary diagnosis in the United States of America; however, relatively few pathology departments in the country have fully implemented an enterprise wide digital pathology system enabled for primary diagnosis. Digital pathology has significant potential to transform pathology practice with several published studies documenting some level of diagnostic equivalence between digital and conventional systems. However, whole slide imaging also has significant potential to disrupt pathology practice, due to the differences in efficiency of manipulating digital images vis-à-vis glass slides, and studies on the efficiency of actual digital pathology workload are lacking. Our randomized, equivalency and efficiency study aimed to replicate clinical workflow, comparing conventional microscopy to a complete digital pathology signout using whole slide images, evaluating the equivalency and efficiency of glass slide to whole slide image reporting, reflective of true pathology practice workloads in the clinical setting. All glass slides representing an entire day’s routine clinical signout workload for six different anatomic pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on Leica Aperio AT2 at ×40 (0.25 µm/pixel). Integration of whole slide images for each accessioned case is through an interface between the Leica eSlide manager database and the laboratory information system, Cerner CoPathPlus. Pathologists utilized a standard institution computer workstation and viewed whole slide images through an internally developed, vendor agnostic whole slide image viewer, named the “MSK Slide Viewer”. Subspecialized pathologists first reported on glass slides from surgical pathology cases using routine clinical workflow. Glass slides were de-identified, scanned, and re-accessioned in the laboratory information system test environment. After a washout period of 13 weeks, pathologists reported the same clinical workload using whole slide image integrated within the laboratory information system. Intraobserver equivalency metrics included top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and the need to order ancillary testing (i.e., recuts, immunohistochemistry). Turnaround time (efficiency) evaluation was defined by the start of each case when opened in the laboratory information system and when the case was completed for that day (i.e., case sent to signout queue or pending ancillary studies). Eight pathologists participated from the following subspecialties: bone and soft tissue, genitourinary, gastrointestinal, breast, gynecologic, and dermatopathology. Glass slides signouts comprised of 204 cases, encompassing 2091 glass slides; and digital signouts comprised of 199 cases, encompassing 2073 whole slide images. The median whole slide image file size was 1.54 GB; scan time/slide, 6 min 24 s; and scan area 32.1 × 18.52 mm. Overall diagnostic equivalency (e.g., top-line diagnosis) was 99.3% between digital and glass slide signout; however, signout using whole slide images showed a median overall 19% decrease in efficiency per case. No significant difference by reader, subspecialty, or specimen type was identified. Our experience is the most comprehensive study to date and shows high intraobserver whole slide image to glass slide equivalence in reporting of true clinical workflows and workloads. Efficiency needs to improve for digital pathology to gain more traction among pathologists.
DA - 2019/02/18/
PY - 2019
DO - 10.1038/s41379-019-0205-0
DP - www.nature.com
VL - 32
SP - 916
EP - 928
LA - En
SN - 1530-0285
ST - Whole slide imaging equivalency and efficiency study
UR - https://www.nature.com/articles/s41379-019-0205-0
Y2 - 2019/02/21/21:15:25
ER -

Whole slide imaging is Food and Drug Administration-approved for primary diagnosis in the United States of America; however, relatively few pathology departments in the country have fully implemented an enterprise wide digital pathology system enabled for primary diagnosis. Digital pathology has significant potential to transform pathology practice with several published studies documenting some level of diagnostic equivalence between digital and conventional systems. However, whole slide imaging also has significant potential to disrupt pathology practice, due to the differences in efficiency of manipulating digital images vis-à-vis glass slides, and studies on the efficiency of actual digital pathology workload are lacking. Our randomized, equivalency and efficiency study aimed to replicate clinical workflow, comparing conventional microscopy to a complete digital pathology signout using whole slide images, evaluating the equivalency and efficiency of glass slide to whole slide image reporting, reflective of true pathology practice workloads in the clinical setting. All glass slides representing an entire day’s routine clinical signout workload for six different anatomic pathology subspecialties at Memorial Sloan Kettering Cancer Center were scanned on Leica Aperio AT2 at ×40 (0.25 µm/pixel). Integration of whole slide images for each accessioned case is through an interface between the Leica eSlide manager database and the laboratory information system, Cerner CoPathPlus. Pathologists utilized a standard institution computer workstation and viewed whole slide images through an internally developed, vendor agnostic whole slide image viewer, named the “MSK Slide Viewer”. Subspecialized pathologists first reported on glass slides from surgical pathology cases using routine clinical workflow. Glass slides were de-identified, scanned, and re-accessioned in the laboratory information system test environment. After a washout period of 13 weeks, pathologists reported the same clinical workload using whole slide image integrated within the laboratory information system. Intraobserver equivalency metrics included top-line diagnosis, margin status, lymphovascular and/or perineural invasion, pathology stage, and the need to order ancillary testing (i.e., recuts, immunohistochemistry). Turnaround time (efficiency) evaluation was defined by the start of each case when opened in the laboratory information system and when the case was completed for that day (i.e., case sent to signout queue or pending ancillary studies). Eight pathologists participated from the following subspecialties: bone and soft tissue, genitourinary, gastrointestinal, breast, gynecologic, and dermatopathology. Glass slides signouts comprised of 204 cases, encompassing 2091 glass slides; and digital signouts comprised of 199 cases, encompassing 2073 whole slide images. The median whole slide image file size was 1.54 GB; scan time/slide, 6 min 24 s; and scan area 32.1 × 18.52 mm. Overall diagnostic equivalency (e.g., top-line diagnosis) was 99.3% between digital and glass slide signout; however, signout using whole slide images showed a median overall 19% decrease in efficiency per case. No significant difference by reader, subspecialty, or specimen type was identified. Our experience is the most comprehensive study to date and shows high intraobserver whole slide image to glass slide equivalence in reporting of true clinical workflows and workloads. Efficiency needs to improve for digital pathology to gain more traction among pathologists.

Clinical-grade computational pathology using weakly supervised deep learning on whole slide images.

Gabriele Campanella, Matthew G. Hanna, Luke Geneslaw, Allen Miraflor, Vitor Werneck Krauss Silva, Klaus J. Busam, Edi Brogi, Victor E. Reuter, David S. Klimstra and Thomas J. Fuchs.

Nat Med, vol. 25, 8, p. 1301-1309, 8/2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{campanella_clinical-grade_2019,
    title = {Clinical-grade computational pathology using weakly supervised deep learning on whole slide images},
    volume = {25},
    issn = {1078-8956, 1546-170X},
    url = {http://www.nature.com/articles/s41591-019-0508-1},
    doi = {10.1038/s41591-019-0508-1},
    language = {en},
    number = {8},
    urldate = {2019-11-26},
    journal = {Nature Medicine},
    author = {Campanella, Gabriele and Hanna, Matthew G. and Geneslaw, Luke and Miraflor, Allen and Werneck Krauss Silva, Vitor and Busam, Klaus J. and Brogi, Edi and Reuter, Victor E. and Klimstra, David S. and Fuchs, Thomas J.},
    month = aug,
    year = {2019},
    pages = {1301--1309},
}

Download Endnote/RIS citation

TY - JOUR
TI - Clinical-grade computational pathology using weakly supervised deep learning on whole slide images
AU - Campanella, Gabriele
AU - Hanna, Matthew G.
AU - Geneslaw, Luke
AU - Miraflor, Allen
AU - Werneck Krauss Silva, Vitor
AU - Busam, Klaus J.
AU - Brogi, Edi
AU - Reuter, Victor E.
AU - Klimstra, David S.
AU - Fuchs, Thomas J.
T2 - Nature Medicine
DA - 2019/08//
PY - 2019
DO - 10.1038/s41591-019-0508-1
DP - DOI.org (Crossref)
VL - 25
IS - 8
SP - 1301
EP - 1309
J2 - Nat Med
LA - en
SN - 1078-8956, 1546-170X
UR - http://www.nature.com/articles/s41591-019-0508-1
Y2 - 2019/11/26/18:16:43
ER -

Comparison of contrast-enhanced and diffusion-weighted MRI in assessment of the terminal ileum in Crohn’s disease patients.

Carl A. J. Puylaert, Jeroen A. W. Tielbeek, Peter J. Schüffler, C. Yung Nio, Karin Horsthuis, Banafsche Mearadji, Cyriel Y. Ponsioen, Frans M. Vos and Jaap Stoker.

Abdominal Radiology, vol. 44, p. 398–405, 2018-8-14

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{puylaert_comparison_2018,
    title = {Comparison of contrast-enhanced and diffusion-weighted {MRI} in assessment of the terminal ileum in {Crohn}’s disease patients},
    volume = {44},
    issn = {2366-004X, 2366-0058},
    url = {http://link.springer.com/10.1007/s00261-018-1734-6},
    doi = {10.1007/s00261-018-1734-6},
    language = {en},
    urldate = {2018-09-04},
    journal = {Abdominal Radiology},
    author = {Puylaert, Carl A. J. and Tielbeek, Jeroen A. W. and Sch\"uffler, Peter J. and Nio, C. Yung and Horsthuis, Karin and Mearadji, Banafsche and Ponsioen, Cyriel Y. and Vos, Frans M. and Stoker, Jaap},
    month = aug,
    year = {2018},
    pages = {398--405},
}

Download Endnote/RIS citation

TY - JOUR
TI - Comparison of contrast-enhanced and diffusion-weighted MRI in assessment of the terminal ileum in Crohn’s disease patients
AU - Puylaert, Carl A. J.
AU - Tielbeek, Jeroen A. W.
AU - Schüffler, Peter J.
AU - Nio, C. Yung
AU - Horsthuis, Karin
AU - Mearadji, Banafsche
AU - Ponsioen, Cyriel Y.
AU - Vos, Frans M.
AU - Stoker, Jaap
T2 - Abdominal Radiology
DA - 2018/08/14/
PY - 2018
DO - 10.1007/s00261-018-1734-6
DP - Crossref
VL - 44
SP - 398
EP - 405
LA - en
SN - 2366-004X, 2366-0058
UR - http://link.springer.com/10.1007/s00261-018-1734-6
Y2 - 2018/09/04/23:18:09
ER -

Terabyte-scale Deep Multiple Instance Learning for Classification and Localization in Pathology.

Gabriele Campanella, Vitor Werneck Krauss Silva and Thomas J. Fuchs.

arXiv:1805.06983 [cs], 2018-05-17

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{campanella_terabyte-scale_2018,
    title = {Terabyte-scale {Deep} {Multiple} {Instance} {Learning} for {Classification} and {Localization} in {Pathology}},
    url = {http://arxiv.org/abs/1805.06983},
    abstract = {In the ﬁeld of computational pathology, the use of decision support systems powered by state-of-the-art deep learning solutions has been hampered by the lack of large labeled datasets. Until recently, studies relied on datasets in the order of few hundreds of slides which are not enough to train a model that can work at scale in the clinic. Here, we have gathered a dataset consisting of 12,160 slides, two orders of magnitude larger than previous datasets in pathology and equivalent to 25 times the pixel count of the entire ImageNet dataset. Given the size of our dataset it is possible for us to train a deep learning model under the Multiple Instance Learning (MIL) assumption where only the overall slide diagnosis is necessary for training, avoiding all the expensive pixelwise annotations that are usually part of supervised learning approaches. We test our framework on a complex task, that of prostate cancer diagnosis on needle biopsies. We performed a thorough evaluation of the performance of our MIL pipeline under several conditions achieving an AUC of 0.98 on a held-out test set of 1,824 slides. These results open the way for training accurate diagnosis prediction models at scale, laying the foundation for decision support system deployment in the clinic.},
    language = {en},
    urldate = {2018-05-21},
    journal = {arXiv:1805.06983 [cs]},
    author = {Campanella, Gabriele and Silva, Vitor Werneck Krauss and Fuchs, Thomas J.},
    month = may,
    year = {2018},
    keywords = {Computer Science - Computer Vision and Pattern Recognition},
}

Download Endnote/RIS citation

TY - JOUR
TI - Terabyte-scale Deep Multiple Instance Learning for Classification and Localization in Pathology
AU - Campanella, Gabriele
AU - Silva, Vitor Werneck Krauss
AU - Fuchs, Thomas J.
T2 - arXiv:1805.06983 [cs]
AB - In the ﬁeld of computational pathology, the use of decision support systems powered by state-of-the-art deep learning solutions has been hampered by the lack of large labeled datasets. Until recently, studies relied on datasets in the order of few hundreds of slides which are not enough to train a model that can work at scale in the clinic. Here, we have gathered a dataset consisting of 12,160 slides, two orders of magnitude larger than previous datasets in pathology and equivalent to 25 times the pixel count of the entire ImageNet dataset. Given the size of our dataset it is possible for us to train a deep learning model under the Multiple Instance Learning (MIL) assumption where only the overall slide diagnosis is necessary for training, avoiding all the expensive pixelwise annotations that are usually part of supervised learning approaches. We test our framework on a complex task, that of prostate cancer diagnosis on needle biopsies. We performed a thorough evaluation of the performance of our MIL pipeline under several conditions achieving an AUC of 0.98 on a held-out test set of 1,824 slides. These results open the way for training accurate diagnosis prediction models at scale, laying the foundation for decision support system deployment in the clinic.
DA - 2018/05/17/
PY - 2018
DP - arXiv.org
LA - en
UR - http://arxiv.org/abs/1805.06983
Y2 - 2018/05/21/17:23:14
KW - Computer Science - Computer Vision and Pattern Recognition
ER -

In the ﬁeld of computational pathology, the use of decision support systems powered by state-of-the-art deep learning solutions has been hampered by the lack of large labeled datasets. Until recently, studies relied on datasets in the order of few hundreds of slides which are not enough to train a model that can work at scale in the clinic. Here, we have gathered a dataset consisting of 12,160 slides, two orders of magnitude larger than previous datasets in pathology and equivalent to 25 times the pixel count of the entire ImageNet dataset. Given the size of our dataset it is possible for us to train a deep learning model under the Multiple Instance Learning (MIL) assumption where only the overall slide diagnosis is necessary for training, avoiding all the expensive pixelwise annotations that are usually part of supervised learning approaches. We test our framework on a complex task, that of prostate cancer diagnosis on needle biopsies. We performed a thorough evaluation of the performance of our MIL pipeline under several conditions achieving an AUC of 0.98 on a held-out test set of 1,824 slides. These results open the way for training accurate diagnosis prediction models at scale, laying the foundation for decision support system deployment in the clinic.

DeepPET: A deep encoder–decoder network for directly solving the PET reconstruction inverse problem.

Ida Häggström, C. Ross Schmidtlein, Gabriele Campanella and Thomas J. Fuchs.

arXiv:1804.0785 [cs.CV], 2018-04-20

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{haggstrom_deeppet:_2018,
    title = {{DeepPET}: {A} deep encoder–decoder network for directly solving the {PET}
    reconstruction inverse problem},
    url = {https://arxiv.org/abs/1804.07851},
    abstract = {Positron emission tomography (PET) is a cornerstone of modern radiology. The ability to detect cancer and metastases in whole body scans fundamentally changed cancer diagnosis and treatment.
    One of the main bottlenecks in the clinical application is the time it takes to reconstruct the anatomical image from the deluge of data in PET
    imaging. State-of-the art methods based on expectation maximization can take hours for a single patient and depend on manual fine-tuning. This
    results not only in financial burden for hospitals but more importantly leads to less efficient patient
    handling, evaluation, and ultimately diagnosis and treatment for patients.
    To overcome this problem we present a novel PET image reconstruction technique based on a deep convolutional encoder–decoder network,
    that takes PET sinogram data as input and directly outputs full PET images. Using realistic simulated data, we demonstrate that our network is able to reconstruct images {\textgreater}100 times faster, and with comparable image quality (in terms of root mean squared error) relative to conventional iterative reconstruction techniques.},
    language = {en},
    journal = {arXiv:1804.0785 [cs.CV]},
    author = {H\"aggstr\"om, Ida and Schmidtlein, C. Ross and Campanella, Gabriele and Fuchs, Thomas J.},
    month = apr,
    year = {2018},
    keywords = {Computer Vision and Pattern Recognition},
}

Download Endnote/RIS citation

TY - JOUR
TI - DeepPET: A deep encoder–decoder network for directly solving the PET
reconstruction inverse problem
AU - Häggström, Ida
AU - Schmidtlein, C. Ross
AU - Campanella, Gabriele
AU - Fuchs, Thomas J.
T2 - arXiv:1804.0785 [cs.CV]
AB - Positron emission tomography (PET) is a cornerstone of modern radiology. The ability to detect cancer and metastases in whole body scans fundamentally changed cancer diagnosis and treatment.
One of the main bottlenecks in the clinical application is the time it takes to reconstruct the anatomical image from the deluge of data in PET
imaging. State-of-the art methods based on expectation maximization can take hours for a single patient and depend on manual fine-tuning. This
results not only in financial burden for hospitals but more importantly leads to less efficient patient
handling, evaluation, and ultimately diagnosis and treatment for patients.
To overcome this problem we present a novel PET image reconstruction technique based on a deep convolutional encoder–decoder network,
that takes PET sinogram data as input and directly outputs full PET images. Using realistic simulated data, we demonstrate that our network is able to reconstruct images >100 times faster, and with comparable image quality (in terms of root mean squared error) relative to conventional iterative reconstruction techniques.
DA - 2018/04/20/
PY - 2018
LA - en
UR - https://arxiv.org/abs/1804.07851
DB - arxiv.org
KW - Computer Vision and Pattern Recognition
ER -

Positron emission tomography (PET) is a cornerstone of modern radiology. The ability to detect cancer and metastases in whole body scans fundamentally changed cancer diagnosis and treatment. One of the main bottlenecks in the clinical application is the time it takes to reconstruct the anatomical image from the deluge of data in PET imaging. State-of-the art methods based on expectation maximization can take hours for a single patient and depend on manual fine-tuning. This results not only in financial burden for hospitals but more importantly leads to less efficient patient handling, evaluation, and ultimately diagnosis and treatment for patients. To overcome this problem we present a novel PET image reconstruction technique based on a deep convolutional encoder–decoder network, that takes PET sinogram data as input and directly outputs full PET images. Using realistic simulated data, we demonstrate that our network is able to reconstruct images >100 times faster, and with comparable image quality (in terms of root mean squared error) relative to conventional iterative reconstruction techniques.

Comprehensive immunohistochemical analysis of PD-L1 shows scarce expression in castration-resistant prostate cancer.

Christian D. Fankhauser, Peter J. Schüffler, Silke Gillessen, Aurelius Omlin, Niels J. Rupp, Jan H. Rueschoff, Thomas Hermanns, Cedric Poyet, Tullio Sulser, Holger Moch and Peter J. Wild.

Oncotarget, vol. 9, 12, 2018-02-13

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{fankhauser_comprehensive_2018,
    title = {Comprehensive immunohistochemical analysis of {PD}-{L1} shows scarce expression in castration-resistant prostate cancer},
    volume = {9},
    issn = {1949-2553},
    url = {http://www.oncotarget.com/fulltext/22888},
    doi = {10.18632/oncotarget.22888},
    language = {en},
    number = {12},
    urldate = {2018-05-31},
    journal = {Oncotarget},
    author = {Fankhauser, Christian D. and Sch\"uffler, Peter J. and Gillessen, Silke and Omlin, Aurelius and Rupp, Niels J. and Rueschoff, Jan H. and Hermanns, Thomas and Poyet, Cedric and Sulser, Tullio and Moch, Holger and Wild, Peter J.},
    month = feb,
    year = {2018},
}

Download Endnote/RIS citation

TY - JOUR
TI - Comprehensive immunohistochemical analysis of PD-L1 shows scarce expression in castration-resistant prostate cancer
AU - Fankhauser, Christian D.
AU - Schüffler, Peter J.
AU - Gillessen, Silke
AU - Omlin, Aurelius
AU - Rupp, Niels J.
AU - Rueschoff, Jan H.
AU - Hermanns, Thomas
AU - Poyet, Cedric
AU - Sulser, Tullio
AU - Moch, Holger
AU - Wild, Peter J.
T2 - Oncotarget
DA - 2018/02/13/
PY - 2018
DO - 10.18632/oncotarget.22888
DP - Crossref
VL - 9
IS - 12
LA - en
SN - 1949-2553
UR - http://www.oncotarget.com/fulltext/22888
Y2 - 2018/05/31/17:41:01
ER -

Towards machine learned quality control: A benchmark for sharpness quantification in digital pathology.

Gabriele Campanella, Arjun R. Rajanna, Lorraine Corsale, Peter J. Schüffler, Yukako Yagi and Thomas J. Fuchs.

Computerized Medical Imaging and Graphics, vol. 65, p. 142-151, 04/2018

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{campanella_towards_2018,
    title = {Towards machine learned quality control: {A} benchmark for sharpness quantification in digital pathology},
    volume = {65},
    issn = {08956111},
    shorttitle = {Towards machine learned quality control},
    url = {https://linkinghub.elsevier.com/retrieve/pii/S0895611117300800},
    doi = {10.1016/j.compmedimag.2017.09.001},
    language = {en},
    urldate = {2019-11-26},
    journal = {Computerized Medical Imaging and Graphics},
    author = {Campanella, Gabriele and Rajanna, Arjun R. and Corsale, Lorraine and Sch\"uffler, Peter J. and Yagi, Yukako and Fuchs, Thomas J.},
    month = apr,
    year = {2018},
    pages = {142--151},
}

Download Endnote/RIS citation

TY - JOUR
TI - Towards machine learned quality control: A benchmark for sharpness quantification in digital pathology
AU - Campanella, Gabriele
AU - Rajanna, Arjun R.
AU - Corsale, Lorraine
AU - Schüffler, Peter J.
AU - Yagi, Yukako
AU - Fuchs, Thomas J.
T2 - Computerized Medical Imaging and Graphics
DA - 2018/04//
PY - 2018
DO - 10.1016/j.compmedimag.2017.09.001
DP - DOI.org (Crossref)
VL - 65
SP - 142
EP - 151
J2 - Computerized Medical Imaging and Graphics
LA - en
SN - 08956111
ST - Towards machine learned quality control
UR - https://linkinghub.elsevier.com/retrieve/pii/S0895611117300800
Y2 - 2019/11/26/18:28:10
ER -

Semiautomatic Assessment of the Terminal Ileum and Colon in Patients with Crohn Disease Using MRI (the VIGOR++ Project).

Carl A.J. Puylaert, Peter J. Schüffler, Robiel E. Naziroglu, Jeroen A.W. Tielbeek, Zhang Li, Jesica C. Makanyanga, Charlotte J. Tutein Nolthenius, C. Yung Nio, Douglas A. Pendsé, Alex Menys, Cyriel Y. Ponsioen, David Atkinson, Alastair Forbes, Joachim M. Buhmann, Thomas J. Fuchs, Haralambos Hatzakis, Lucas J. van Vliet, Jaap Stoker, Stuart A. Taylor and Frans M. Vos.

Academic Radiology, vol. 25, 8, p. 1038-1045, 2/2018

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{puylaert_semiautomatic_2018,
    title = {Semiautomatic {Assessment} of the {Terminal} {Ileum} and {Colon} in {Patients} with {Crohn} {Disease} {Using} {MRI} (the {VIGOR}++ {Project})},
    volume = {25},
    issn = {10766332},
    url = {http://linkinghub.elsevier.com/retrieve/pii/S1076633218300060},
    doi = {10.1016/j.acra.2017.12.024},
    language = {en},
    number = {8},
    urldate = {2018-05-21},
    journal = {Academic Radiology},
    author = {Puylaert, Carl A.J. and Sch\"uffler, Peter J. and Naziroglu, Robiel E. and Tielbeek, Jeroen A.W. and Li, Zhang and Makanyanga, Jesica C. and Tutein Nolthenius, Charlotte J. and Nio, C. Yung and Pends\'e, Douglas A. and Menys, Alex and Ponsioen, Cyriel Y. and Atkinson, David and Forbes, Alastair and Buhmann, Joachim M. and Fuchs, Thomas J. and Hatzakis, Haralambos and van Vliet, Lucas J. and Stoker, Jaap and Taylor, Stuart A. and Vos, Frans M.},
    month = feb,
    year = {2018},
    pages = {1038--1045},
}

Download Endnote/RIS citation

TY - JOUR
TI - Semiautomatic Assessment of the Terminal Ileum and Colon in Patients with Crohn Disease Using MRI (the VIGOR++ Project)
AU - Puylaert, Carl A.J.
AU - Schüffler, Peter J.
AU - Naziroglu, Robiel E.
AU - Tielbeek, Jeroen A.W.
AU - Li, Zhang
AU - Makanyanga, Jesica C.
AU - Tutein Nolthenius, Charlotte J.
AU - Nio, C. Yung
AU - Pendsé, Douglas A.
AU - Menys, Alex
AU - Ponsioen, Cyriel Y.
AU - Atkinson, David
AU - Forbes, Alastair
AU - Buhmann, Joachim M.
AU - Fuchs, Thomas J.
AU - Hatzakis, Haralambos
AU - van Vliet, Lucas J.
AU - Stoker, Jaap
AU - Taylor, Stuart A.
AU - Vos, Frans M.
T2 - Academic Radiology
DA - 2018/02//
PY - 2018
DO - 10.1016/j.acra.2017.12.024
DP - Crossref
VL - 25
IS - 8
SP - 1038
EP - 1045
LA - en
SN - 10766332
UR - http://linkinghub.elsevier.com/retrieve/pii/S1076633218300060
Y2 - 2018/05/21/12:24:37
ER -

Computational Pathology.

Peter J. Schüffler, Qing Zhong, Peter J. Wild and Thomas J. Fuchs.

In: Johannes Haybäck (ed.) Mechanisms of Molecular Carcinogenesis - Volume 2, 1st ed. 2017 edition, Springer, ISBN 3-319-53660-5, 21. Jun. 2017

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{schuffler_computational_2017,
    edition = {1st ed. 2017 edition},
    title = {Computational {Pathology}},
    isbn = {3-319-53660-5},
    url = {http://www.springer.com/de/book/9783319536606},
    booktitle = {Mechanisms of {Molecular} {Carcinogenesis} - {Volume} 2},
    publisher = {Springer},
    author = {Sch\"uffler, Peter J. and Zhong, Qing and Wild, Peter J. and Fuchs, Thomas J.},
    editor = {Hayb\"ack, Johannes},
    month = jun,
    year = {2017},
}

Download Endnote/RIS citation

TY - CHAP
TI - Computational Pathology
AU - Schüffler, Peter J.
AU - Zhong, Qing
AU - Wild, Peter J.
AU - Fuchs, Thomas J.
T2 - Mechanisms of Molecular Carcinogenesis - Volume 2
A2 - Haybäck, Johannes
DA - 2017/06/21/
PY - 2017
ET - 1st ed. 2017 edition
PB - Springer
SN - 3-319-53660-5
UR - http://www.springer.com/de/book/9783319536606
ER -

Scaling of cytoskeletal organization with cell size in Drosophila.

Alison K. Spencer, Andrew J. Schaumberg and Jennifer A. Zallen.

Mol. Biol. Cell, p. mbc.E16-10-0691, 2017-04-12

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{spencer_scaling_2017,
    title = {Scaling of cytoskeletal organization with cell size in {Drosophila}},
    issn = {1059-1524, 1939-4586},
    url = {http://www.molbiolcell.org/content/early/2017/04/10/mbc.E16-10-0691},
    doi = {10.1091/mbc.E16-10-0691},
    abstract = {Spatially organized macromolecular complexes are essential for cell and tissue function, but the mechanisms that organize micron-scale structures within cells are not well understood. Microtubule-based structures such as mitotic spindles scale with cell size, but less is known about the scaling of actin structures within cells. Actin-rich denticle precursors cover the ventral surface of the Drosophila embryo and larva and provide templates for cuticular structures involved in larval locomotion. Using quantitative imaging and statistical modeling, we demonstrate that denticle number and spacing scale with cell size over a wide range of cell lengths in embryos and larvae. Denticle number and spacing are reduced under space-limited conditions, and both features robustly scale over a ten-fold increase in cell length during larval growth. We show that the relationship between cell length and denticle spacing can be recapitulated by specific mathematical equations in embryos and larvae, and that accurate denticle spacing requires an intact microtubule network and the microtubule minus-end-binding protein, Patronin. These results identify a novel mechanism of microtubule-dependent actin scaling that maintains precise patterns of actin organization during tissue growth.},
    language = {en},
    urldate = {2017-04-24},
    journal = {Molecular Biology of the Cell},
    author = {Spencer, Alison K. and Schaumberg, Andrew J. and Zallen, Jennifer A.},
    month = apr,
    year = {2017},
    pages = {mbc.E16--10--0691},
}

Download Endnote/RIS citation

TY - JOUR
TI - Scaling of cytoskeletal organization with cell size in Drosophila
AU - Spencer, Alison K.
AU - Schaumberg, Andrew J.
AU - Zallen, Jennifer A.
T2 - Molecular Biology of the Cell
AB - Spatially organized macromolecular complexes are essential for cell and tissue function, but the mechanisms that organize micron-scale structures within cells are not well understood. Microtubule-based structures such as mitotic spindles scale with cell size, but less is known about the scaling of actin structures within cells. Actin-rich denticle precursors cover the ventral surface of the Drosophila embryo and larva and provide templates for cuticular structures involved in larval locomotion. Using quantitative imaging and statistical modeling, we demonstrate that denticle number and spacing scale with cell size over a wide range of cell lengths in embryos and larvae. Denticle number and spacing are reduced under space-limited conditions, and both features robustly scale over a ten-fold increase in cell length during larval growth. We show that the relationship between cell length and denticle spacing can be recapitulated by specific mathematical equations in embryos and larvae, and that accurate denticle spacing requires an intact microtubule network and the microtubule minus-end-binding protein, Patronin. These results identify a novel mechanism of microtubule-dependent actin scaling that maintains precise patterns of actin organization during tissue growth.
DA - 2017/04/12/
PY - 2017
DO - 10.1091/mbc.E16-10-0691
DP - www.molbiolcell.org
SP - mbc.E16
EP - 10-0691
J2 - Mol. Biol. Cell
LA - en
SN - 1059-1524, 1939-4586
UR - http://www.molbiolcell.org/content/early/2017/04/10/mbc.E16-10-0691
Y2 - 2017/04/24/16:19:42
ER -

Spatially organized macromolecular complexes are essential for cell and tissue function, but the mechanisms that organize micron-scale structures within cells are not well understood. Microtubule-based structures such as mitotic spindles scale with cell size, but less is known about the scaling of actin structures within cells. Actin-rich denticle precursors cover the ventral surface of the Drosophila embryo and larva and provide templates for cuticular structures involved in larval locomotion. Using quantitative imaging and statistical modeling, we demonstrate that denticle number and spacing scale with cell size over a wide range of cell lengths in embryos and larvae. Denticle number and spacing are reduced under space-limited conditions, and both features robustly scale over a ten-fold increase in cell length during larval growth. We show that the relationship between cell length and denticle spacing can be recapitulated by specific mathematical equations in embryos and larvae, and that accurate denticle spacing requires an intact microtubule network and the microtubule minus-end-binding protein, Patronin. These results identify a novel mechanism of microtubule-dependent actin scaling that maintains precise patterns of actin organization during tissue growth.

Hybrid Deep Learning on Single Wide-field Optical Coherence Tomography Scans Accurately Classifies Glaucoma Suspects.

Hassan Muhammad, Thomas J. Fuchs, Nicole De Cuir, Carlos G. De Moraes, Dana M. Blumberg, Jeffrey M. Liebmann, Robert Ritch and Donald C. Hood.

Journal of Glaucoma, vol. 26, 12, 10/2017

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{muhammad_hybrid_2017,
    title = {Hybrid {Deep} {Learning} on {Single} {Wide}-field {Optical} {Coherence} {Tomography} {Scans} {Accurately} {Classifies} {Glaucoma} {Suspects}.},
    volume = {26},
    issn = {1057-0829},
    shorttitle = {Hybrid {Deep} {Learning} on {Single} {Wide}-field {Optical} {Coherence} {Tomography} {Scans} {Accurately} {Classifies} {Glaucoma} {Suspects}},
    url = {http://Insights.ovid.com/crossref?an=00061198-900000000-98595},
    doi = {10.1097/IJG.0000000000000765},
    language = {en},
    number = {12},
    urldate = {2017-11-20},
    journal = {Journal of Glaucoma},
    author = {Muhammad, Hassan and Fuchs, Thomas J. and De Cuir, Nicole and De Moraes, Carlos G. and Blumberg, Dana M. and Liebmann, Jeffrey M. and Ritch, Robert and Hood, Donald C.},
    month = oct,
    year = {2017},
}

Download Endnote/RIS citation

TY - JOUR
TI - Hybrid Deep Learning on Single Wide-field Optical Coherence Tomography Scans Accurately Classifies Glaucoma Suspects.
AU - Muhammad, Hassan
AU - Fuchs, Thomas J.
AU - De Cuir, Nicole
AU - De Moraes, Carlos G.
AU - Blumberg, Dana M.
AU - Liebmann, Jeffrey M.
AU - Ritch, Robert
AU - Hood, Donald C.
T2 - Journal of Glaucoma
DA - 2017/10//
PY - 2017
DO - 10.1097/IJG.0000000000000765
DP - CrossRef
VL - 26
IS - 12
LA - en
SN - 1057-0829
ST - Hybrid Deep Learning on Single Wide-field Optical Coherence Tomography Scans Accurately Classifies Glaucoma Suspects
UR - http://Insights.ovid.com/crossref?an=00061198-900000000-98595
Y2 - 2017/11/20/15:25:19
ER -

MRI-Based Surgical Planning for Lumbar Spinal Stenosis.

Gabriele Abbati, Stefan Bauer, Sebastian Winklhofer, Peter J. Schüffler, Ulrike Held, Jakob M. Burgstaller, Johann Steurer and Joachim M. Buhmann.

In: Maxime Descoteaux, Lena Maier-Hein, Alfred Franz, Pierre Jannin, D. Louis Collins and Simon Duchesne (eds.) Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017, vol. 10435, p. 116-124, Lecture Notes in Computer Science, Springer, ISBN 978-3-319-66179-7, 2017

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{descoteaux_mri-based_2017,
    address = {Cham},
    series = {Lecture {Notes} in {Computer} {Science}},
    title = {{MRI}-{Based} {Surgical} {Planning} for {Lumbar} {Spinal} {Stenosis}},
    volume = {10435},
    isbn = {978-3-319-66179-7},
    url = {http://link.springer.com/10.1007/978-3-319-66179-7_14},
    urldate = {2017-09-18},
    booktitle = {Medical {Image} {Computing} and {Computer}-{Assisted} {Intervention} − {MICCAI} 2017},
    publisher = {Springer},
    author = {Abbati, Gabriele and Bauer, Stefan and Winklhofer, Sebastian and Sch\"uffler, Peter J. and Held, Ulrike and Burgstaller, Jakob M. and Steurer, Johann and Buhmann, Joachim M.},
    editor = {Descoteaux, Maxime and Maier-Hein, Lena and Franz, Alfred and Jannin, Pierre and Collins, D. Louis and Duchesne, Simon},
    year = {2017},
    doi = {10.1007/978-3-319-66179-7_14},
    pages = {116--124},
}

Download Endnote/RIS citation

TY - CHAP
TI - MRI-Based Surgical Planning for Lumbar Spinal Stenosis
AU - Abbati, Gabriele
AU - Bauer, Stefan
AU - Winklhofer, Sebastian
AU - Schüffler, Peter J.
AU - Held, Ulrike
AU - Burgstaller, Jakob M.
AU - Steurer, Johann
AU - Buhmann, Joachim M.
T2 - Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017
A2 - Descoteaux, Maxime
A2 - Maier-Hein, Lena
A2 - Franz, Alfred
A2 - Jannin, Pierre
A2 - Collins, D. Louis
A2 - Duchesne, Simon
T3 - Lecture Notes in Computer Science
CY - Cham
DA - 2017///
PY - 2017
DP - CrossRef
VL - 10435
SP - 116
EP - 124
PB - Springer
SN - 978-3-319-66179-7
UR - http://link.springer.com/10.1007/978-3-319-66179-7_14
Y2 - 2017/09/18/12:44:49
ER -

Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations.

Peter J. Schüffler, Judy Sarungbam, Hassan Muhammad, Ed Reznik, Satish K. Tickoo and Thomas J. Fuchs.

Proceedings of the 1st Machine Learning for Healthcare Conference, Machine Learning for Healthcare, vol. 56, p. 191-208, Proceedings of Machine Learning Research, PMLR, 18 Aug 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schuffler_mitochondria-based_2016,
    address = {Los Angeles},
    series = {Proceedings of {Machine} {Learning} {Research}},
    title = {Mitochondria-based {Renal} {Cell} {Carcinoma} {Subtyping}: {Learning} from {Deep} vs. {Flat} {Feature} {Representations}},
    volume = {56},
    url = {http://proceedings.mlr.press/v56/Schuffler16.html},
    abstract = {Accurate subtyping of renal cell carcinoma (RCC) is of crucial importance for understanding disease progression and for making informed treatment decisions. New discoveries of significant mitochondria alterations between subtypes makes immunohistochemical (IHC) staining based image classification an imperative.
    Until now, accurate quantification and subtyping was made impossible by huge IHC variations, the absence of cell membrane staining for cytoplasm segmentation as well as the complete lack of systems for robust and reproducible image based classification.
    In this paper we present a comprehensive classification framework to overcome these challenges for tissue microarrays (TMA) of RCCs. We compare and evaluate models based on domain specific hand-crafted "flat"-features versus "deep" feature representations from various layers of a pre-trained convolutional neural network (CNN).
    The best model reaches a cross-validation accuracy of 89\%, which demonstrates for the first time, that robust mitochondria-based subtyping of renal cancer is feasible.},
    language = {English},
    booktitle = {Proceedings of the 1st {Machine} {Learning} for {Healthcare} {Conference}},
    publisher = {PMLR},
    author = {Sch\"uffler, Peter J. and Sarungbam, Judy and Muhammad, Hassan and Reznik, Ed and Tickoo, Satish K. and Fuchs, Thomas J.},
    editor = {Finale, Doshi-Valez and Fackler, Jim and Kale, David and Wallace, Byron and Weins, Jenna},
    month = aug,
    year = {2016},
    pages = {191--208},
}

Download Endnote/RIS citation

TY - CONF
TI - Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations
AU - Schüffler, Peter J.
AU - Sarungbam, Judy
AU - Muhammad, Hassan
AU - Reznik, Ed
AU - Tickoo, Satish K.
AU - Fuchs, Thomas J.
T2 - Machine Learning for Healthcare
A2 - Finale, Doshi-Valez
A2 - Fackler, Jim
A2 - Kale, David
A2 - Wallace, Byron
A2 - Weins, Jenna
T3 - Proceedings of Machine Learning Research
AB - Accurate subtyping of renal cell carcinoma (RCC) is of crucial importance for understanding disease progression and for making informed treatment decisions. New discoveries of significant mitochondria alterations between subtypes makes immunohistochemical (IHC) staining based image classification an imperative.
Until now, accurate quantification and subtyping was made impossible by huge IHC variations, the absence of cell membrane staining for cytoplasm segmentation as well as the complete lack of systems for robust and reproducible image based classification.
In this paper we present a comprehensive classification framework to overcome these challenges for tissue microarrays (TMA) of RCCs. We compare and evaluate models based on domain specific hand-crafted "flat"-features versus "deep" feature representations from various layers of a pre-trained convolutional neural network (CNN).
The best model reaches a cross-validation accuracy of 89%, which demonstrates for the first time, that robust mitochondria-based subtyping of renal cancer is feasible.
C1 - Los Angeles
C3 - Proceedings of the 1st Machine Learning for Healthcare Conference
DA - 2016/08/18/
PY - 2016
VL - 56
SP - 191
EP - 208
LA - English
PB - PMLR
UR - http://proceedings.mlr.press/v56/Schuffler16.html
ER -

Accurate subtyping of renal cell carcinoma (RCC) is of crucial importance for understanding disease progression and for making informed treatment decisions. New discoveries of significant mitochondria alterations between subtypes makes immunohistochemical (IHC) staining based image classification an imperative. Until now, accurate quantification and subtyping was made impossible by huge IHC variations, the absence of cell membrane staining for cytoplasm segmentation as well as the complete lack of systems for robust and reproducible image based classification. In this paper we present a comprehensive classification framework to overcome these challenges for tissue microarrays (TMA) of RCCs. We compare and evaluate models based on domain specific hand-crafted "flat"-features versus "deep" feature representations from various layers of a pre-trained convolutional neural network (CNN). The best model reaches a cross-validation accuracy of 89%, which demonstrates for the first time, that robust mitochondria-based subtyping of renal cancer is feasible.

A generalized computer vision approach to mapping crop fields in heterogeneous agricultural landscapes.

Stephanie R. Debats, Dee Luo, Lyndon D. Estes, Thomas J. Fuchs and Kelly K. Caylor.

Remote Sensing of Environment, vol. 179, p. 210-221, June 15, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{debats_generalized_2016,
    title = {A generalized computer vision approach to mapping crop fields in heterogeneous agricultural landscapes},
    volume = {179},
    issn = {0034-4257},
    url = {http://www.sciencedirect.com/science/article/pii/S0034425716301031},
    doi = {10.1016/j.rse.2016.03.010},
    abstract = {Smallholder farms dominate in many parts of the world, particularly Sub-Saharan Africa. These systems are characterized by small, heterogeneous, and often indistinct field patterns, requiring a specialized methodology to map agricultural land cover. Using a variety of sites in South Africa, we present a new approach to mapping agricultural fields, based on efficient extraction of a vast set of simple, highly correlated, and interdependent features, followed by a random forest classifier. We achieved similar high performance across agricultural types, including the spectrally indistinct smallholder fields as well as the more easily distinguishable commercial fields, and demonstrated the ability to generalize performance across large geographic areas. In sensitivity analyses, we determined multi-temporal information provided greater gains in performance than the addition of multi-spectral bands available in DigitalGlobe Worldview-2 imagery.},
    urldate = {2017-05-19},
    journal = {Remote Sensing of Environment},
    author = {Debats, Stephanie R. and Luo, Dee and Estes, Lyndon D. and Fuchs, Thomas J. and Caylor, Kelly K.},
    month = jun,
    year = {2016},
    keywords = {Agriculture, Land cover, Sub-Saharan Africa, computer vision, machine learning},
    pages = {210--221},
}

Download Endnote/RIS citation

TY - JOUR
TI - A generalized computer vision approach to mapping crop fields in heterogeneous agricultural landscapes
AU - Debats, Stephanie R.
AU - Luo, Dee
AU - Estes, Lyndon D.
AU - Fuchs, Thomas J.
AU - Caylor, Kelly K.
T2 - Remote Sensing of Environment
AB - Smallholder farms dominate in many parts of the world, particularly Sub-Saharan Africa. These systems are characterized by small, heterogeneous, and often indistinct field patterns, requiring a specialized methodology to map agricultural land cover. Using a variety of sites in South Africa, we present a new approach to mapping agricultural fields, based on efficient extraction of a vast set of simple, highly correlated, and interdependent features, followed by a random forest classifier. We achieved similar high performance across agricultural types, including the spectrally indistinct smallholder fields as well as the more easily distinguishable commercial fields, and demonstrated the ability to generalize performance across large geographic areas. In sensitivity analyses, we determined multi-temporal information provided greater gains in performance than the addition of multi-spectral bands available in DigitalGlobe Worldview-2 imagery.
DA - 2016/06/15/
PY - 2016
DO - 10.1016/j.rse.2016.03.010
DP - ScienceDirect
VL - 179
SP - 210
EP - 221
J2 - Remote Sensing of Environment
SN - 0034-4257
UR - http://www.sciencedirect.com/science/article/pii/S0034425716301031
Y2 - 2017/05/19/10:44:55
KW - Agriculture
KW - Land cover
KW - Sub-Saharan Africa
KW - computer vision
KW - machine learning
ER -

Smallholder farms dominate in many parts of the world, particularly Sub-Saharan Africa. These systems are characterized by small, heterogeneous, and often indistinct field patterns, requiring a specialized methodology to map agricultural land cover. Using a variety of sites in South Africa, we present a new approach to mapping agricultural fields, based on efficient extraction of a vast set of simple, highly correlated, and interdependent features, followed by a random forest classifier. We achieved similar high performance across agricultural types, including the spectrally indistinct smallholder fields as well as the more easily distinguishable commercial fields, and demonstrated the ability to generalize performance across large geographic areas. In sensitivity analyses, we determined multi-temporal information provided greater gains in performance than the addition of multi-spectral bands available in DigitalGlobe Worldview-2 imagery.

Is There an Association Between Pain and Magnetic Resonance Imaging Parameters in Patients With Lumbar Spinal Stenosis?

Jakob M. Burgstaller, Peter J. Schüffler, Joachim M. Buhmann, Gustav Andreisek, Sebastian Winklhofer, Filippo Del Grande, Michèle Mattle, Florian Brunner, Georgios Karakoumis, Johann Steurer and Ulrike Held.

Spine, vol. 41, 17, p. 1053-1062, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{burgstaller_is_2016,
    title = {Is {There} an {Association} {Between} {Pain} and {Magnetic} {Resonance} {Imaging} {Parameters} in {Patients} {With} {Lumbar} {Spinal} {Stenosis}?},
    volume = {41},
    issn = {0362-2436},
    shorttitle = {Is {There} an {Association} {Between} {Pain} and {Magnetic} {Resonance} {Imaging} {Parameters} in {Patients} {With} {Lumbar} {Spinal} {Stenosis}?},
    url = {http://Insights.ovid.com/crossref?an=00007632-201609010-00015},
    doi = {10.1097/BRS.0000000000001544},
    abstract = {STUDY DESIGN:
    A prospective multicenter cohort study.
    OBJECTIVE:
    The aim of this study was to identify an association between pain and magnetic resonance imaging (MRI) parameters in patients with lumbar spinal stenosis (LSS).
    SUMMARY OF BACKGROUND DATA:
    At present, the relationship between abnormal MRI findings and pain in patients with LSS is still unclear.
    METHODS:
    First, we conducted a systematic literature search. We identified relationships of relevant MRI parameters and pain in patients with LSS. Second, we addressed the study question with a thorough descriptive and graphical analysis to establish a relationship between MRI parameters and pain using data of the LSS outcome study (LSOS).
    RESULTS:
    In the systematic review including four papers about the associations between radiological findings in the MRI and pain, the authors of two articles reported no association and two of them did. Of the latters, only one study found a moderate correlation between leg pain measured by Visual Analog Scale (VAS) and the degree of stenosis assessed by spine surgeons. In the data of the LSOS study, we could not identify a relevant association between any of the MRI parameters and buttock, leg, and back pain, quantified by the Spinal Stenosis Measure (SSM) and the Numeric Rating Scale (NRS). Even by restricting the analysis to the level of the lumbar spine with the most prominent radiological "stenosis," no relevant association could be shown.
    CONCLUSION:
    Despite a thorough analysis of the data, we were not able to prove any correlation between radiological findings (MRI) and the severity of pain. There is a need for innovative "methods/techniques" to learn more about the causal relationship between radiological findings and the patients' pain-related complaints.
    LEVEL OF EVIDENCE:
    2.},
    language = {en},
    number = {17},
    urldate = {2017-02-11},
    journal = {Spine},
    author = {Burgstaller, Jakob M. and Sch\"uffler, Peter J. and Buhmann, Joachim M. and Andreisek, Gustav and Winklhofer, Sebastian and Del Grande, Filippo and Mattle, Mich\`ele and Brunner, Florian and Karakoumis, Georgios and Steurer, Johann and Held, Ulrike},
    year = {2016},
    pages = {1053--1062},
}

Download Endnote/RIS citation

TY - JOUR
TI - Is There an Association Between Pain and Magnetic Resonance Imaging Parameters in Patients With Lumbar Spinal Stenosis?
AU - Burgstaller, Jakob M.
AU - Schüffler, Peter J.
AU - Buhmann, Joachim M.
AU - Andreisek, Gustav
AU - Winklhofer, Sebastian
AU - Del Grande, Filippo
AU - Mattle, Michèle
AU - Brunner, Florian
AU - Karakoumis, Georgios
AU - Steurer, Johann
AU - Held, Ulrike
T2 - Spine
AB - STUDY DESIGN:
A prospective multicenter cohort study.
OBJECTIVE:
The aim of this study was to identify an association between pain and magnetic resonance imaging (MRI) parameters in patients with lumbar spinal stenosis (LSS).
SUMMARY OF BACKGROUND DATA:
At present, the relationship between abnormal MRI findings and pain in patients with LSS is still unclear.
METHODS:
First, we conducted a systematic literature search. We identified relationships of relevant MRI parameters and pain in patients with LSS. Second, we addressed the study question with a thorough descriptive and graphical analysis to establish a relationship between MRI parameters and pain using data of the LSS outcome study (LSOS).
RESULTS:
In the systematic review including four papers about the associations between radiological findings in the MRI and pain, the authors of two articles reported no association and two of them did. Of the latters, only one study found a moderate correlation between leg pain measured by Visual Analog Scale (VAS) and the degree of stenosis assessed by spine surgeons. In the data of the LSOS study, we could not identify a relevant association between any of the MRI parameters and buttock, leg, and back pain, quantified by the Spinal Stenosis Measure (SSM) and the Numeric Rating Scale (NRS). Even by restricting the analysis to the level of the lumbar spine with the most prominent radiological "stenosis," no relevant association could be shown.
CONCLUSION:
Despite a thorough analysis of the data, we were not able to prove any correlation between radiological findings (MRI) and the severity of pain. There is a need for innovative "methods/techniques" to learn more about the causal relationship between radiological findings and the patients' pain-related complaints.
LEVEL OF EVIDENCE:
2.
DA - 2016///
PY - 2016
DO - 10.1097/BRS.0000000000001544
DP - CrossRef
VL - 41
IS - 17
SP - 1053
EP - 1062
LA - en
SN - 0362-2436
ST - Is There an Association Between Pain and Magnetic Resonance Imaging Parameters in Patients With Lumbar Spinal Stenosis?
UR - http://Insights.ovid.com/crossref?an=00007632-201609010-00015
Y2 - 2017/02/11/00:39:09
ER -

STUDY DESIGN: A prospective multicenter cohort study. OBJECTIVE: The aim of this study was to identify an association between pain and magnetic resonance imaging (MRI) parameters in patients with lumbar spinal stenosis (LSS). SUMMARY OF BACKGROUND DATA: At present, the relationship between abnormal MRI findings and pain in patients with LSS is still unclear. METHODS: First, we conducted a systematic literature search. We identified relationships of relevant MRI parameters and pain in patients with LSS. Second, we addressed the study question with a thorough descriptive and graphical analysis to establish a relationship between MRI parameters and pain using data of the LSS outcome study (LSOS). RESULTS: In the systematic review including four papers about the associations between radiological findings in the MRI and pain, the authors of two articles reported no association and two of them did. Of the latters, only one study found a moderate correlation between leg pain measured by Visual Analog Scale (VAS) and the degree of stenosis assessed by spine surgeons. In the data of the LSOS study, we could not identify a relevant association between any of the MRI parameters and buttock, leg, and back pain, quantified by the Spinal Stenosis Measure (SSM) and the Numeric Rating Scale (NRS). Even by restricting the analysis to the level of the lumbar spine with the most prominent radiological "stenosis," no relevant association could be shown. CONCLUSION: Despite a thorough analysis of the data, we were not able to prove any correlation between radiological findings (MRI) and the severity of pain. There is a need for innovative "methods/techniques" to learn more about the causal relationship between radiological findings and the patients' pain-related complaints. LEVEL OF EVIDENCE: 2.

DeepScope: Nonintrusive Whole Slide Saliency Annotation and Prediction from Pathologists at the Microscope.

Andrew J. Schaumberg, S. Joseph Sirintrapun, Hikmat A. Al-Ahmadie, Peter J. Schüffler and Thomas J. Fuchs.

13th International Conference on Computational Intelligence methods for Bioinformatics and Biostatistics, CIBB, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schaumberg_deepscope:_2016,
    address = {Stirling, United Kingdom},
    title = {{DeepScope}: {Nonintrusive} {Whole} {Slide} {Saliency} {Annotation} and {Prediction} from {Pathologists} at the {Microscope}},
    shorttitle = {{DeepScope}},
    url = {http://www.cs.stir.ac.uk/events/cibb2016/},
    abstract = {Modern digital pathology departments have grown to produce whole-slide image data at petabyte scale, an unprecedented treasure chest for medical machine learning tasks.
    Unfortunately, most digital slides are not annotated at the image level, hindering large-scale application of supervised learning. Manual labeling is prohibitive, requiring pathologists with decades of training and outstanding clinical service responsibilities. This problem is further aggravated by the United States Food and Drug Administration's ruling that primary diagnosis must come from a glass slide rather than a digital image.
    We present the first end-to-end framework to overcome this problem, gathering annotations in a nonintrusive manner during a pathologist's routine clinical work: (i) microscope-specific 3D-printed commodity camera mounts are used to video record the glass-slide-based clinical diagnosis process; (ii) after routine scanning of the whole slide, the video frames are registered to the digital slide; (iii) motion and observing time are estimated to generate a spatial and temporal saliency map of the whole slide.
    Demonstrating the utility of these annotations, we train a convolutional neural network that detects diagnosis-relevant salient regions, then report accuracy of 85.15\% in bladder and 91.50\% in prostate, with 75.00\% accuracy when training on prostate but predicting in bladder, despite different pathologists examining the different tissues.},
    language = {English},
    booktitle = {13th {International} {Conference} on {Computational} {Intelligence} methods for {Bioinformatics} and {Biostatistics}},
    author = {Schaumberg, Andrew J. and Sirintrapun, S. Joseph and Al-Ahmadie, Hikmat A. and Sch\"uffler, Peter J. and Fuchs, Thomas J.},
    year = {2016},
}

Download Endnote/RIS citation

TY - CONF
TI - DeepScope: Nonintrusive Whole Slide Saliency Annotation and Prediction from Pathologists at the Microscope
AU - Schaumberg, Andrew J.
AU - Sirintrapun, S. Joseph
AU - Al-Ahmadie, Hikmat A.
AU - Schüffler, Peter J.
AU - Fuchs, Thomas J.
T2 - CIBB
AB - Modern digital pathology departments have grown to produce whole-slide image data at petabyte scale, an unprecedented treasure chest for medical machine learning tasks.
Unfortunately, most digital slides are not annotated at the image level, hindering large-scale application of supervised learning. Manual labeling is prohibitive, requiring pathologists with decades of training and outstanding clinical service responsibilities. This problem is further aggravated by the United States Food and Drug Administration's ruling that primary diagnosis must come from a glass slide rather than a digital image.
We present the first end-to-end framework to overcome this problem, gathering annotations in a nonintrusive manner during a pathologist's routine clinical work: (i) microscope-specific 3D-printed commodity camera mounts are used to video record the glass-slide-based clinical diagnosis process; (ii) after routine scanning of the whole slide, the video frames are registered to the digital slide; (iii) motion and observing time are estimated to generate a spatial and temporal saliency map of the whole slide.
Demonstrating the utility of these annotations, we train a convolutional neural network that detects diagnosis-relevant salient regions, then report accuracy of 85.15% in bladder and 91.50% in prostate, with 75.00% accuracy when training on prostate but predicting in bladder, despite different pathologists examining the different tissues.
C1 - Stirling, United Kingdom
C3 - 13th International Conference on Computational Intelligence methods for Bioinformatics and Biostatistics
DA - 2016///
PY - 2016
LA - English
ST - DeepScope
UR - http://www.cs.stir.ac.uk/events/cibb2016/
ER -

Modern digital pathology departments have grown to produce whole-slide image data at petabyte scale, an unprecedented treasure chest for medical machine learning tasks. Unfortunately, most digital slides are not annotated at the image level, hindering large-scale application of supervised learning. Manual labeling is prohibitive, requiring pathologists with decades of training and outstanding clinical service responsibilities. This problem is further aggravated by the United States Food and Drug Administration's ruling that primary diagnosis must come from a glass slide rather than a digital image. We present the first end-to-end framework to overcome this problem, gathering annotations in a nonintrusive manner during a pathologist's routine clinical work: (i) microscope-specific 3D-printed commodity camera mounts are used to video record the glass-slide-based clinical diagnosis process; (ii) after routine scanning of the whole slide, the video frames are registered to the digital slide; (iii) motion and observing time are estimated to generate a spatial and temporal saliency map of the whole slide. Demonstrating the utility of these annotations, we train a convolutional neural network that detects diagnosis-relevant salient regions, then report accuracy of 85.15% in bladder and 91.50% in prostate, with 75.00% accuracy when training on prostate but predicting in bladder, despite different pathologists examining the different tissues.

Big Data in der Medizin.

Qing Zhong, Roman Barnert, Gunnar Ratsch, Thomas J. Fuchs and Peter J. Wild.

Leading Opinions: Hämatologie & Onkologie, p. 102-105, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{zhong_big_2016,
    title = {Big {Data} in der {Medizin}},
    abstract = {IT-Systeme in Krankenh\"ausern bereiten Unmengen von Daten aus unterschiedlichen Quellen, die von spezialisierten interdisziplin\"aren Teams gewonnen worden sind, zu wertvollen Informationen auf. Durch die Ausweitung der Analysen auf mehrere Tausend anonymisierte Patienten und die Vernetzung mit anderen Kliniken und Forschungszentren wird ein \"Okosystem zum Wohle des Patienten geschaffen. Dieses auf Big-Data-Ans\"atzen gest\"utzte, evidenzbasierte und patientenorientierte Gesundheitswesen ist in naher Zukunft Realit\"at.},
    language = {German},
    journal = {Leading Opinions: H\"amatologie \& Onkologie},
    author = {Zhong, Qing and Barnert, Roman and Ratsch, Gunnar and Fuchs, Thomas J. and Wild, Peter J.},
    year = {2016},
    pages = {102--105},
}

Download Endnote/RIS citation

TY - NEWS
TI - Big Data in der Medizin
AU - Zhong, Qing
AU - Barnert, Roman
AU - Ratsch, Gunnar
AU - Fuchs, Thomas J.
AU - Wild, Peter J.
T2 - Leading Opinions: Hämatologie & Onkologie
AB - IT-Systeme in Krankenhäusern bereiten Unmengen von Daten aus unterschiedlichen Quellen, die von spezialisierten interdisziplinären Teams gewonnen worden sind, zu wertvollen Informationen auf. Durch die Ausweitung der Analysen auf mehrere Tausend anonymisierte Patienten und die Vernetzung mit anderen Kliniken und Forschungszentren wird ein Ökosystem zum Wohle des Patienten geschaffen. Dieses auf Big-Data-Ansätzen gestützte, evidenzbasierte und patientenorientierte Gesundheitswesen ist in naher Zukunft Realität.
DA - 2016///
PY - 2016
SP - 102
EP - 105
LA - German
ER -

IT-Systeme in Krankenhäusern bereiten Unmengen von Daten aus unterschiedlichen Quellen, die von spezialisierten interdisziplinären Teams gewonnen worden sind, zu wertvollen Informationen auf. Durch die Ausweitung der Analysen auf mehrere Tausend anonymisierte Patienten und die Vernetzung mit anderen Kliniken und Forschungszentren wird ein Ökosystem zum Wohle des Patienten geschaffen. Dieses auf Big-Data-Ansätzen gestützte, evidenzbasierte und patientenorientierte Gesundheitswesen ist in naher Zukunft Realität.

Cancer-secreted AGR2 induces programmed cell death in normal cells.

Elizabeth A. Vitello, Sue-Ing Quek, Heather Kincaid, Thomas Fuchs, Daniel J. Crichton, Pamela Troisch and Alvin Y. Liu.

Oncotarget, vol. 5, 0, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{vitello_cancer-secreted_2016,
    title = {Cancer-secreted {AGR2} induces programmed cell death in normal cells},
    volume = {5},
    issn = {1949-2553},
    url = {http://www.impactjournals.com/oncotarget/index.php?journal=oncotarget&page=article&op=view&path%5B%5D=9921},
    abstract = {Anterior Gradient 2 (AGR2) is a protein expressed in many solid tumor types including prostate, pancreatic, breast and lung. AGR2 functions as a protein disulfide isomerase in the endoplasmic reticulum. However, AGR2 is secreted by cancer cells that overexpress this molecule. Secretion of AGR2 was also found in salamander limb regeneration. Due to its ubiquity, tumor secretion of AGR2 must serve an important role in cancer, yet its molecular function is largely unknown. This study examined the effect of cancer-secreted AGR2 on normal cells. Prostate stromal cells were cultured, and tissue digestion media containing AGR2 prepared from prostate primary cancer 10-076 CP and adenocarcinoma LuCaP 70CR xenograft were added. The control were tissue digestion media containing no AGR2 prepared from benign prostate 10-076 NP and small cell carcinoma LuCaP 145.1 xenograft. In the presence of tumor-secreted AGR2, the stromal cells were found to undergo programmed cell death (PCD) characterized by formation of cellular blebs, cell shrinkage, and DNA fragmentation as seen when the stromal cells were UV irradiated or treated by a pro-apoptotic drug. PCD could be prevented with the addition of the monoclonal AGR2-neutralizing antibody P3A5. DNA microarray analysis of LuCaP 70CR media-treated vs . LuCaP 145.1 media-treated cells showed downregulation of the gene SAT1 as a major change in cells exposed to AGR2. RT-PCR analysis confirmed the array result. SAT1 encodes spermidine/spermine N 1 -acetyltransferase, which maintains intracellular polyamine levels. Abnormal polyamine metabolism as a result of altered SAT1 activity has an adverse effect on cells through the induction of PCD.},
    number = {0},
    urldate = {2016-06-16},
    journal = {Oncotarget},
    author = {Vitello, Elizabeth A. and Quek, Sue-Ing and Kincaid, Heather and Fuchs, Thomas and Crichton, Daniel J. and Troisch, Pamela and Liu, Alvin Y.},
    year = {2016},
}

Download Endnote/RIS citation

TY - JOUR
TI - Cancer-secreted AGR2 induces programmed cell death in normal cells
AU - Vitello, Elizabeth A.
AU - Quek, Sue-Ing
AU - Kincaid, Heather
AU - Fuchs, Thomas
AU - Crichton, Daniel J.
AU - Troisch, Pamela
AU - Liu, Alvin Y.
T2 - Oncotarget
AB - Anterior Gradient 2 (AGR2) is a protein expressed in many solid tumor types including prostate, pancreatic, breast and lung. AGR2 functions as a protein disulfide isomerase in the endoplasmic reticulum. However, AGR2 is secreted by cancer cells that overexpress this molecule. Secretion of AGR2 was also found in salamander limb regeneration. Due to its ubiquity, tumor secretion of AGR2 must serve an important role in cancer, yet its molecular function is largely unknown. This study examined the effect of cancer-secreted AGR2 on normal cells. Prostate stromal cells were cultured, and tissue digestion media containing AGR2 prepared from prostate primary cancer 10-076 CP and adenocarcinoma LuCaP 70CR xenograft were added. The control were tissue digestion media containing no AGR2 prepared from benign prostate 10-076 NP and small cell carcinoma LuCaP 145.1 xenograft. In the presence of tumor-secreted AGR2, the stromal cells were found to undergo programmed cell death (PCD) characterized by formation of cellular blebs, cell shrinkage, and DNA fragmentation as seen when the stromal cells were UV irradiated or treated by a pro-apoptotic drug. PCD could be prevented with the addition of the monoclonal AGR2-neutralizing antibody P3A5. DNA microarray analysis of LuCaP 70CR media-treated vs . LuCaP 145.1 media-treated cells showed downregulation of the gene SAT1 as a major change in cells exposed to AGR2. RT-PCR analysis confirmed the array result. SAT1 encodes spermidine/spermine N 1 -acetyltransferase, which maintains intracellular polyamine levels. Abnormal polyamine metabolism as a result of altered SAT1 activity has an adverse effect on cells through the induction of PCD.
DA - 2016///
PY - 2016
DP - www.impactjournals.com
VL - 5
IS - 0
SN - 1949-2553
UR - http://www.impactjournals.com/oncotarget/index.php?journal=oncotarget&page=article&op=view&path%5B%5D=9921
Y2 - 2016/06/16/02:00:19
ER -

Anterior Gradient 2 (AGR2) is a protein expressed in many solid tumor types including prostate, pancreatic, breast and lung. AGR2 functions as a protein disulfide isomerase in the endoplasmic reticulum. However, AGR2 is secreted by cancer cells that overexpress this molecule. Secretion of AGR2 was also found in salamander limb regeneration. Due to its ubiquity, tumor secretion of AGR2 must serve an important role in cancer, yet its molecular function is largely unknown. This study examined the effect of cancer-secreted AGR2 on normal cells. Prostate stromal cells were cultured, and tissue digestion media containing AGR2 prepared from prostate primary cancer 10-076 CP and adenocarcinoma LuCaP 70CR xenograft were added. The control were tissue digestion media containing no AGR2 prepared from benign prostate 10-076 NP and small cell carcinoma LuCaP 145.1 xenograft. In the presence of tumor-secreted AGR2, the stromal cells were found to undergo programmed cell death (PCD) characterized by formation of cellular blebs, cell shrinkage, and DNA fragmentation as seen when the stromal cells were UV irradiated or treated by a pro-apoptotic drug. PCD could be prevented with the addition of the monoclonal AGR2-neutralizing antibody P3A5. DNA microarray analysis of LuCaP 70CR media-treated vs . LuCaP 145.1 media-treated cells showed downregulation of the gene SAT1 as a major change in cells exposed to AGR2. RT-PCR analysis confirmed the array result. SAT1 encodes spermidine/spermine N 1 -acetyltransferase, which maintains intracellular polyamine levels. Abnormal polyamine metabolism as a result of altered SAT1 activity has an adverse effect on cells through the induction of PCD.

Multi-Organ Cancer Classification and Survival Analysis.

Stefan Bauer, Nicolas Carion, Peter Schüffler, Thomas Fuchs, Peter Wild and Joachim M. Buhmann.

arXiv:1606.00897 [cs, q-bio, stat], 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{bauer_multi-organ_2016,
    title = {Multi-{Organ} {Cancer} {Classification} and {Survival} {Analysis}},
    url = {http://arxiv.org/abs/1606.00897},
    abstract = {Accurate and robust cell nuclei classification is the cornerstone for a wider range of tasks in digital and Computational Pathology. However, most machine learning systems require extensive labeling from expert pathologists for each individual problem at hand, with no or limited abilities for knowledge transfer between datasets and organ sites. In this paper we implement and evaluate a variety of deep neural network models and model ensembles for nuclei classification in renal cell cancer (RCC) and prostate cancer (PCa). We propose a convolutional neural network system based on residual learning which significantly improves over the state-of-the-art in cell nuclei classification. However, the main thrust of our work is to demonstrate for the first time the models ability to transfer the learned concepts from one organ type to another, with robust performance. Finally, we show that the combination of tissue types during training increases not only classification accuracy but also overall survival analysis. The best model, trained on combined data of RCC and PCa, exhibits optimal performance on PCa classification and better survival group stratification than an expert pathologist (\$p=0.006\$). All code, image data and expert labels are made publicly available to serve as benchmark for the community for future research into computational pathology.},
    urldate = {2016-06-16},
    journal = {arXiv:1606.00897 [cs, q-bio, stat]},
    author = {Bauer, Stefan and Carion, Nicolas and Sch\"uffler, Peter and Fuchs, Thomas and Wild, Peter and Buhmann, Joachim M.},
    year = {2016},
    keywords = {Computer Science - Learning, Quantitative Biology - Quantitative Methods, Quantitative Biology - Tissues and Organs, Statistics - Machine Learning},
}

Download Endnote/RIS citation

TY - JOUR
TI - Multi-Organ Cancer Classification and Survival Analysis
AU - Bauer, Stefan
AU - Carion, Nicolas
AU - Schüffler, Peter
AU - Fuchs, Thomas
AU - Wild, Peter
AU - Buhmann, Joachim M.
T2 - arXiv:1606.00897 [cs, q-bio, stat]
AB - Accurate and robust cell nuclei classification is the cornerstone for a wider range of tasks in digital and Computational Pathology. However, most machine learning systems require extensive labeling from expert pathologists for each individual problem at hand, with no or limited abilities for knowledge transfer between datasets and organ sites. In this paper we implement and evaluate a variety of deep neural network models and model ensembles for nuclei classification in renal cell cancer (RCC) and prostate cancer (PCa). We propose a convolutional neural network system based on residual learning which significantly improves over the state-of-the-art in cell nuclei classification. However, the main thrust of our work is to demonstrate for the first time the models ability to transfer the learned concepts from one organ type to another, with robust performance. Finally, we show that the combination of tissue types during training increases not only classification accuracy but also overall survival analysis. The best model, trained on combined data of RCC and PCa, exhibits optimal performance on PCa classification and better survival group stratification than an expert pathologist ($p=0.006$). All code, image data and expert labels are made publicly available to serve as benchmark for the community for future research into computational pathology.
DA - 2016///
PY - 2016
DP - arXiv.org
UR - http://arxiv.org/abs/1606.00897
Y2 - 2016/06/16/01:52:59
KW - Computer Science - Learning
KW - Quantitative Biology - Quantitative Methods
KW - Quantitative Biology - Tissues and Organs
KW - Statistics - Machine Learning
ER -

Accurate and robust cell nuclei classification is the cornerstone for a wider range of tasks in digital and Computational Pathology. However, most machine learning systems require extensive labeling from expert pathologists for each individual problem at hand, with no or limited abilities for knowledge transfer between datasets and organ sites. In this paper we implement and evaluate a variety of deep neural network models and model ensembles for nuclei classification in renal cell cancer (RCC) and prostate cancer (PCa). We propose a convolutional neural network system based on residual learning which significantly improves over the state-of-the-art in cell nuclei classification. However, the main thrust of our work is to demonstrate for the first time the models ability to transfer the learned concepts from one organ type to another, with robust performance. Finally, we show that the combination of tissue types during training increases not only classification accuracy but also overall survival analysis. The best model, trained on combined data of RCC and PCa, exhibits optimal performance on PCa classification and better survival group stratification than an expert pathologist ($p=0.006$). All code, image data and expert labels are made publicly available to serve as benchmark for the community for future research into computational pathology.

Classifying Renal Cell Carcinoma by Using Convolutional Neural Networks to Deconstruct Pathological Images.

Hassan Muhammad, Peter J. Schüffler, Judy Sarungbam, Satish K. Tickoo and Thomas Fuchs.

Vincent du Vigneaud Memorial Research Symposium, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@misc{muhammad_classifying_2016,
    address = {Weill Cornell Medicine},
    type = {Poster},
    title = {Classifying {Renal} {Cell} {Carcinoma} by {Using} {Convolutional} {Neural} {Networks} to {Deconstruct} {Pathological} {Images}.},
    author = {Muhammad, Hassan},
    collaborator = {Sch\"uffler, Peter J. and Sarungbam, Judy and Tickoo, Satish K. and Fuchs, Thomas},
    year = {2016},
}

Download Endnote/RIS citation

TY - SLIDE
TI - Classifying Renal Cell Carcinoma by Using Convolutional Neural Networks to Deconstruct Pathological Images.
T2 - Vincent du Vigneaud Memorial Research Symposium
A2 - Muhammad, Hassan
CY - Weill Cornell Medicine
DA - 2016///
PY - 2016
M3 - Poster
ER -

Image-Based Computational Quantification and Visualization of Genetic Alterations and Tumour Heterogeneity.

Qing Zhong, Jan H. Rüschoff, Tiannan Guo, Maria Gabrani, Peter J. Schüffler, Markus Rechsteiner, Yansheng Liu, Thomas J. Fuchs, Niels J. Rupp, Christian Fankhauser, Joachim M. Buhmann, Sven Perner, Cédric Poyet, Miriam Blattner, Davide Soldini, Holger Moch, Mark A. Rubin, Aurelia Noske, Josef Rüschoff, Michael C. Haffner, Wolfram Jochum and Peter J. Wild.

Scientific Reports, vol. 6, p. 24146, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{zhong_image-based_2016,
    title = {Image-{Based} {Computational} {Quantification} and {Visualization} of {Genetic} {Alterations} and {Tumour} {Heterogeneity}},
    volume = {6},
    issn = {2045-2322},
    url = {http://www.nature.com/articles/srep24146},
    doi = {10.1038/srep24146},
    urldate = {2016-04-12},
    journal = {Scientific Reports},
    author = {Zhong, Qing and R\"uschoff, Jan H. and Guo, Tiannan and Gabrani, Maria and Sch\"uffler, Peter J. and Rechsteiner, Markus and Liu, Yansheng and Fuchs, Thomas J. and Rupp, Niels J. and Fankhauser, Christian and Buhmann, Joachim M. and Perner, Sven and Poyet, C\'edric and Blattner, Miriam and Soldini, Davide and Moch, Holger and Rubin, Mark A. and Noske, Aurelia and R\"uschoff, Josef and Haffner, Michael C. and Jochum, Wolfram and Wild, Peter J.},
    year = {2016},
    pages = {24146},
}

Download Endnote/RIS citation

TY - JOUR
TI - Image-Based Computational Quantification and Visualization of Genetic Alterations and Tumour Heterogeneity
AU - Zhong, Qing
AU - Rüschoff, Jan H.
AU - Guo, Tiannan
AU - Gabrani, Maria
AU - Schüffler, Peter J.
AU - Rechsteiner, Markus
AU - Liu, Yansheng
AU - Fuchs, Thomas J.
AU - Rupp, Niels J.
AU - Fankhauser, Christian
AU - Buhmann, Joachim M.
AU - Perner, Sven
AU - Poyet, Cédric
AU - Blattner, Miriam
AU - Soldini, Davide
AU - Moch, Holger
AU - Rubin, Mark A.
AU - Noske, Aurelia
AU - Rüschoff, Josef
AU - Haffner, Michael C.
AU - Jochum, Wolfram
AU - Wild, Peter J.
T2 - Scientific Reports
DA - 2016///
PY - 2016
DO - 10.1038/srep24146
DP - CrossRef
VL - 6
SP - 24146
SN - 2045-2322
UR - http://www.nature.com/articles/srep24146
Y2 - 2016/04/12/01:49:30
ER -

Autonomous Terrain Classification With Co- and Self-Training Approach.

K. Otsu, M. Ono, T. J. Fuchs, I. Baldwin and T. Kubota.

IEEE Robotics and Automation Letters, vol. 1, 2, p. 814-819, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{otsu_autonomous_2016,
    title = {Autonomous {Terrain} {Classification} {With} {Co}- and {Self}-{Training} {Approach}},
    volume = {1},
    issn = {2377-3766},
    url = {http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true&arnumber=7397920},
    doi = {10.1109/LRA.2016.2525040},
    abstract = {Identifying terrain type is crucial to safely operating planetary exploration rovers. Vision-based terrain classifiers, which are typically trained by thousands of labeled images using machine learning methods, have proven to be particularly successful. However, since planetary rovers are to boldly go where no one has gone before, training data are usually not available a priori; instead, rovers have to quickly learn from their own experiences in an early phase of surface operation. This research addresses the challenge by combining two key ideas. The first idea is to use both onboard imagery and vibration data, and let rovers learn from physical experiences through self-supervised learning. The underlying fact is that visually similar terrain may be disambiguated by mechanical vibrations. The second idea is to employ the co- and self-training approaches. The idea of co-training is to train two classifiers separately for vision and vibration data, and re-train them iteratively on each other's output. Meanwhile, the self-training approach, applied only to the vision-based classifier, re-trains the classifier on its own output. Both approaches essentially increase the amount of labels, hence enable the terrain classifiers to operate from a sparse training dataset. The proposed approach was validated with a four-wheeled test rover in Mars-analogous terrain, including bedrock, soil, and sand. The co-training setup based on support vector machines with color and wavelet-based features successfully estimated terrain types with 82\% accuracy with only three labeled images.},
    number = {2},
    journal = {IEEE Robotics and Automation Letters},
    author = {Otsu, K. and Ono, M. and Fuchs, T. J. and Baldwin, I. and Kubota, T.},
    year = {2016},
    keywords = {Image color analysis, Mars, Mars-analogous terrain, Semantic Scene Understanding, Soil, Space Robotics, Support vector machines, Training, Training data, Visual Learning, Wheels, aerospace computing, autonomous terrain classification, co-training, color features, four-wheeled test rover, image classification, image colour analysis, learning (artificial intelligence), mechanical vibrations, onboard imagery, planetary rovers, self-supervised learning, self-training, terrain mapping, vibration data, vibrations, vision-based classifier, wavelet transforms, wavelet-based features},
    pages = {814--819},
}

Download Endnote/RIS citation

TY - JOUR
TI - Autonomous Terrain Classification With Co- and Self-Training Approach
AU - Otsu, K.
AU - Ono, M.
AU - Fuchs, T. J.
AU - Baldwin, I.
AU - Kubota, T.
T2 - IEEE Robotics and Automation Letters
AB - Identifying terrain type is crucial to safely operating planetary exploration rovers. Vision-based terrain classifiers, which are typically trained by thousands of labeled images using machine learning methods, have proven to be particularly successful. However, since planetary rovers are to boldly go where no one has gone before, training data are usually not available a priori; instead, rovers have to quickly learn from their own experiences in an early phase of surface operation. This research addresses the challenge by combining two key ideas. The first idea is to use both onboard imagery and vibration data, and let rovers learn from physical experiences through self-supervised learning. The underlying fact is that visually similar terrain may be disambiguated by mechanical vibrations. The second idea is to employ the co- and self-training approaches. The idea of co-training is to train two classifiers separately for vision and vibration data, and re-train them iteratively on each other's output. Meanwhile, the self-training approach, applied only to the vision-based classifier, re-trains the classifier on its own output. Both approaches essentially increase the amount of labels, hence enable the terrain classifiers to operate from a sparse training dataset. The proposed approach was validated with a four-wheeled test rover in Mars-analogous terrain, including bedrock, soil, and sand. The co-training setup based on support vector machines with color and wavelet-based features successfully estimated terrain types with 82% accuracy with only three labeled images.
DA - 2016///
PY - 2016
DO - 10.1109/LRA.2016.2525040
DP - IEEE Xplore
VL - 1
IS - 2
SP - 814
EP - 819
SN - 2377-3766
UR - http://ieeexplore.ieee.org/xpl/articleDetails.jsp?reload=true&arnumber=7397920
KW - Image color analysis
KW - Mars
KW - Mars-analogous terrain
KW - Semantic Scene Understanding
KW - Soil
KW - Space Robotics
KW - Support vector machines
KW - Training
KW - Training data
KW - Visual Learning
KW - Wheels
KW - aerospace computing
KW - autonomous terrain classification
KW - co-training
KW - color features
KW - four-wheeled test rover
KW - image classification
KW - image colour analysis
KW - learning (artificial intelligence)
KW - mechanical vibrations
KW - onboard imagery
KW - planetary rovers
KW - self-supervised learning
KW - self-training
KW - terrain mapping
KW - vibration data
KW - vibrations
KW - vision-based classifier
KW - wavelet transforms
KW - wavelet-based features
ER -

Identifying terrain type is crucial to safely operating planetary exploration rovers. Vision-based terrain classifiers, which are typically trained by thousands of labeled images using machine learning methods, have proven to be particularly successful. However, since planetary rovers are to boldly go where no one has gone before, training data are usually not available a priori; instead, rovers have to quickly learn from their own experiences in an early phase of surface operation. This research addresses the challenge by combining two key ideas. The first idea is to use both onboard imagery and vibration data, and let rovers learn from physical experiences through self-supervised learning. The underlying fact is that visually similar terrain may be disambiguated by mechanical vibrations. The second idea is to employ the co- and self-training approaches. The idea of co-training is to train two classifiers separately for vision and vibration data, and re-train them iteratively on each other's output. Meanwhile, the self-training approach, applied only to the vision-based classifier, re-trains the classifier on its own output. Both approaches essentially increase the amount of labels, hence enable the terrain classifiers to operate from a sparse training dataset. The proposed approach was validated with a four-wheeled test rover in Mars-analogous terrain, including bedrock, soil, and sand. The co-training setup based on support vector machines with color and wavelet-based features successfully estimated terrain types with 82% accuracy with only three labeled images.

Predicting Drug Interactions and Mutagenicity with Ensemble Classifiers on Subgraphs of Molecules.

Andrew Schaumberg, Angela Yu, Tatsuhiro Koshi, Xiaochan Zong and Santoshkalyan Rayadhurgam.

arXiv:1601.07233 [cs, stat], 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schaumberg_predicting_2016,
    title = {Predicting {Drug} {Interactions} and {Mutagenicity} with {Ensemble} {Classifiers} on {Subgraphs} of {Molecules}},
    url = {http://arxiv.org/abs/1601.07233},
    abstract = {In this study, we intend to solve a mutual information problem in interacting molecules of any type, such as proteins, nucleic acids, and small molecules. Using machine learning techniques, we accurately predict pairwise interactions, which can be of medical and biological importance. Graphs are are useful in this problem for their generality to all types of molecules, due to the inherent association of atoms through atomic bonds. Subgraphs can represent different molecular domains. These domains can be biologically significant as most molecules only have portions that are of functional significance and can interact with other domains. Thus, we use subgraphs as features in different machine learning algorithms to predict if two drugs interact and predict potential single molecule effects.},
    urldate = {2016-01-28},
    journal = {arXiv:1601.07233 [cs, stat]},
    author = {Schaumberg, Andrew and Yu, Angela and Koshi, Tatsuhiro and Zong, Xiaochan and Rayadhurgam, Santoshkalyan},
    year = {2016},
    keywords = {Computer Science - Learning, I.2.1, J.3, Statistics - Machine Learning},
}

Download Endnote/RIS citation

TY - JOUR
TI - Predicting Drug Interactions and Mutagenicity with Ensemble Classifiers on Subgraphs of Molecules
AU - Schaumberg, Andrew
AU - Yu, Angela
AU - Koshi, Tatsuhiro
AU - Zong, Xiaochan
AU - Rayadhurgam, Santoshkalyan
T2 - arXiv:1601.07233 [cs, stat]
AB - In this study, we intend to solve a mutual information problem in interacting molecules of any type, such as proteins, nucleic acids, and small molecules. Using machine learning techniques, we accurately predict pairwise interactions, which can be of medical and biological importance. Graphs are are useful in this problem for their generality to all types of molecules, due to the inherent association of atoms through atomic bonds. Subgraphs can represent different molecular domains. These domains can be biologically significant as most molecules only have portions that are of functional significance and can interact with other domains. Thus, we use subgraphs as features in different machine learning algorithms to predict if two drugs interact and predict potential single molecule effects.
DA - 2016///
PY - 2016
DP - arXiv.org
UR - http://arxiv.org/abs/1601.07233
Y2 - 2016/01/28/19:24:38
KW - Computer Science - Learning
KW - I.2.1
KW - J.3
KW - Statistics - Machine Learning
ER -

In this study, we intend to solve a mutual information problem in interacting molecules of any type, such as proteins, nucleic acids, and small molecules. Using machine learning techniques, we accurately predict pairwise interactions, which can be of medical and biological importance. Graphs are are useful in this problem for their generality to all types of molecules, due to the inherent association of atoms through atomic bonds. Subgraphs can represent different molecular domains. These domains can be biologically significant as most molecules only have portions that are of functional significance and can interact with other domains. Thus, we use subgraphs as features in different machine learning algorithms to predict if two drugs interact and predict potential single molecule effects.

Real-Time Data Mining of Massive Data Streams from Synoptic Sky Surveys.

S. G. Djorgovski, M. J. Graham, C. Donalek, A. A. Mahabal, A. J. Drake, M. Turmon and T.J. Fuchs.

Future Generation Computer Systems, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{djorgovski_real-time_2016,
    title = {Real-{Time} {Data} {Mining} of {Massive} {Data} {Streams} from {Synoptic} {Sky} {Surveys}},
    issn = {0167-739X},
    url = {http://www.sciencedirect.com/science/article/pii/S0167739X1500326X},
    doi = {10.1016/j.future.2015.10.013},
    abstract = {The nature of scientific and technological data collection is evolving rapidly: data volumes and rates grow exponentially, with increasing complexity and information content, and there has been a transition from static data sets to data streams that must be analyzed in real time. Interesting or anomalous phenomena must be quickly characterized and followed up with additional measurements via optimal deployment of limited assets. Modern astronomy presents a variety of such phenomena in the form of transient events in digital synoptic sky surveys, including cosmic explosions (supernovae, gamma ray bursts), relativistic phenomena (black hole formation, jets), potentially hazardous asteroids, etc. We have been developing a set of machine learning tools to detect, classify and plan a response to transient events for astronomy applications, using the Catalina Real-time Transient Survey (CRTS) as a scientific and methodological testbed. The ability to respond rapidly to the potentially most interesting events is a key bottleneck that limits the scientific returns from the current and anticipated synoptic sky surveys. Similar challenge arise in other contexts, from environmental monitoring using sensor networks to autonomous spacecraft systems. Given the exponential growth of data rates, and the time-critical response, we need a fully automated and robust approach. We describe the results obtained to date, and the possible future developments.},
    urldate = {2016-01-20},
    journal = {Future Generation Computer Systems},
    author = {Djorgovski, S. G. and Graham, M. J. and Donalek, C. and Mahabal, A. A. and Drake, A. J. and Turmon, M. and Fuchs, T.J.},
    year = {2016},
    keywords = {Automated decision making, Bayesian methods, Massive data streams, Sky surveys, machine learning},
}

Download Endnote/RIS citation

TY - JOUR
TI - Real-Time Data Mining of Massive Data Streams from Synoptic Sky Surveys
AU - Djorgovski, S. G.
AU - Graham, M. J.
AU - Donalek, C.
AU - Mahabal, A. A.
AU - Drake, A. J.
AU - Turmon, M.
AU - Fuchs, T.J.
T2 - Future Generation Computer Systems
AB - The nature of scientific and technological data collection is evolving rapidly: data volumes and rates grow exponentially, with increasing complexity and information content, and there has been a transition from static data sets to data streams that must be analyzed in real time. Interesting or anomalous phenomena must be quickly characterized and followed up with additional measurements via optimal deployment of limited assets. Modern astronomy presents a variety of such phenomena in the form of transient events in digital synoptic sky surveys, including cosmic explosions (supernovae, gamma ray bursts), relativistic phenomena (black hole formation, jets), potentially hazardous asteroids, etc. We have been developing a set of machine learning tools to detect, classify and plan a response to transient events for astronomy applications, using the Catalina Real-time Transient Survey (CRTS) as a scientific and methodological testbed. The ability to respond rapidly to the potentially most interesting events is a key bottleneck that limits the scientific returns from the current and anticipated synoptic sky surveys. Similar challenge arise in other contexts, from environmental monitoring using sensor networks to autonomous spacecraft systems. Given the exponential growth of data rates, and the time-critical response, we need a fully automated and robust approach. We describe the results obtained to date, and the possible future developments.
DA - 2016///
PY - 2016
DO - 10.1016/j.future.2015.10.013
DP - ScienceDirect
J2 - Future Generation Computer Systems
SN - 0167-739X
UR - http://www.sciencedirect.com/science/article/pii/S0167739X1500326X
Y2 - 2016/01/20/14:32:18
KW - Automated decision making
KW - Bayesian methods
KW - Massive data streams
KW - Sky surveys
KW - machine learning
ER -

The nature of scientific and technological data collection is evolving rapidly: data volumes and rates grow exponentially, with increasing complexity and information content, and there has been a transition from static data sets to data streams that must be analyzed in real time. Interesting or anomalous phenomena must be quickly characterized and followed up with additional measurements via optimal deployment of limited assets. Modern astronomy presents a variety of such phenomena in the form of transient events in digital synoptic sky surveys, including cosmic explosions (supernovae, gamma ray bursts), relativistic phenomena (black hole formation, jets), potentially hazardous asteroids, etc. We have been developing a set of machine learning tools to detect, classify and plan a response to transient events for astronomy applications, using the Catalina Real-time Transient Survey (CRTS) as a scientific and methodological testbed. The ability to respond rapidly to the potentially most interesting events is a key bottleneck that limits the scientific returns from the current and anticipated synoptic sky surveys. Similar challenge arise in other contexts, from environmental monitoring using sensor networks to autonomous spacecraft systems. Given the exponential growth of data rates, and the time-critical response, we need a fully automated and robust approach. We describe the results obtained to date, and the possible future developments.

Automatic single cell segmentation on highly multiplexed tissue images.

Peter J. Schüffler, Denis Schapiro, Charlotte Giesen, Hao A. O. Wang, Bernd Bodenmiller and Joachim M. Buhmann.

Cytometry Part A, vol. 87, 10, p. 936-942, 10/2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_automatic_2015,
    title = {Automatic single cell segmentation on highly multiplexed tissue images},
    volume = {87},
    issn = {15524922},
    shorttitle = {Automatic single cell segmentation on highly multiplexed tissue images},
    url = {http://doi.wiley.com/10.1002/cyto.a.22702},
    doi = {10.1002/cyto.a.22702},
    abstract = {The combination of mass cytometry and immunohistochemistry (IHC) enables new histopathological imaging methods in which dozens of proteins and protein modifications can be visualized simultaneously in a single tissue section. The power of multiplexing combined with spatial information and quantification was recently illustrated on breast cancer tissue and was described as next-generation IHC. Robust, accurate, and high-throughput cell segmentation is crucial for the analysis of this new generation of IHC data. To this end, we propose a watershed-based cell segmentation, which uses a nuclear marker and multiple membrane markers, the latter automatically selected based on their correlation. In comparison with the state-of-the-art segmentation pipelines, which are only using a single marker for object detection, we could show that the use of multiple markers can significantly increase the segmentation power, and thus, multiplexed information should be used and not ignored during the segmentation. Furthermore, we provide a novel, user-friendly open-source toolbox for the automatic segmentation of multiplexed histopathological images. (c) 2015 International Society for Advancement of Cytometry.},
    language = {en},
    number = {10},
    urldate = {2017-02-14},
    journal = {Cytometry Part A},
    author = {Sch\"uffler, Peter J. and Schapiro, Denis and Giesen, Charlotte and Wang, Hao A. O. and Bodenmiller, Bernd and Buhmann, Joachim M.},
    month = oct,
    year = {2015},
    pages = {936--942},
}

Download Endnote/RIS citation

TY - JOUR
TI - Automatic single cell segmentation on highly multiplexed tissue images
AU - Schüffler, Peter J.
AU - Schapiro, Denis
AU - Giesen, Charlotte
AU - Wang, Hao A. O.
AU - Bodenmiller, Bernd
AU - Buhmann, Joachim M.
T2 - Cytometry Part A
AB - The combination of mass cytometry and immunohistochemistry (IHC) enables new histopathological imaging methods in which dozens of proteins and protein modifications can be visualized simultaneously in a single tissue section. The power of multiplexing combined with spatial information and quantification was recently illustrated on breast cancer tissue and was described as next-generation IHC. Robust, accurate, and high-throughput cell segmentation is crucial for the analysis of this new generation of IHC data. To this end, we propose a watershed-based cell segmentation, which uses a nuclear marker and multiple membrane markers, the latter automatically selected based on their correlation. In comparison with the state-of-the-art segmentation pipelines, which are only using a single marker for object detection, we could show that the use of multiple markers can significantly increase the segmentation power, and thus, multiplexed information should be used and not ignored during the segmentation. Furthermore, we provide a novel, user-friendly open-source toolbox for the automatic segmentation of multiplexed histopathological images. (c) 2015 International Society for Advancement of Cytometry.
DA - 2015/10//
PY - 2015
DO - 10.1002/cyto.a.22702
DP - CrossRef
VL - 87
IS - 10
SP - 936
EP - 942
LA - en
SN - 15524922
ST - Automatic single cell segmentation on highly multiplexed tissue images
UR - http://doi.wiley.com/10.1002/cyto.a.22702
Y2 - 2017/02/14/19:42:24
ER -

The combination of mass cytometry and immunohistochemistry (IHC) enables new histopathological imaging methods in which dozens of proteins and protein modifications can be visualized simultaneously in a single tissue section. The power of multiplexing combined with spatial information and quantification was recently illustrated on breast cancer tissue and was described as next-generation IHC. Robust, accurate, and high-throughput cell segmentation is crucial for the analysis of this new generation of IHC data. To this end, we propose a watershed-based cell segmentation, which uses a nuclear marker and multiple membrane markers, the latter automatically selected based on their correlation. In comparison with the state-of-the-art segmentation pipelines, which are only using a single marker for object detection, we could show that the use of multiple markers can significantly increase the segmentation power, and thus, multiplexed information should be used and not ignored during the segmentation. Furthermore, we provide a novel, user-friendly open-source toolbox for the automatic segmentation of multiplexed histopathological images. (c) 2015 International Society for Advancement of Cytometry.

Understanding Neural Networks Through Deep Visualization.

Jason Yosinski, Jeff Clune, Anh Nguyen, Thomas Fuchs and Hod Lipson.

ICML Deep Learning Workshop, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{yosinski_understanding_2015,
    title = {Understanding {Neural} {Networks} {Through} {Deep} {Visualization}},
    url = {http://arxiv.org/abs/1506.06579},
    abstract = {Recent years have produced great advances in training large, deep neural networks (DNNs), including notable successes in training convolutional neural networks (convnets) to recognize natural images. However, our understanding of how these models work, especially what computations they perform at intermediate layers, has lagged behind. Progress in the field will be further accelerated by the development of better tools for visualizing and interpreting neural nets. We introduce two such tools here. The first is a tool that visualizes the activations produced on each layer of a trained convnet as it processes an image or video (e.g. a live webcam stream). We have found that looking at live activations that change in response to user input helps build valuable intuitions about how convnets work. The second tool enables visualizing features at each layer of a DNN via regularized optimization in image space. Because previous versions of this idea produced less recognizable images, here we introduce several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations. Both tools are open source and work on a pre-trained convnet with minimal setup.},
    urldate = {2015-12-03},
    booktitle = {{ICML} {Deep} {Learning} {Workshop}},
    author = {Yosinski, Jason and Clune, Jeff and Nguyen, Anh and Fuchs, Thomas and Lipson, Hod},
    year = {2015},
}

Download Endnote/RIS citation

TY - CONF
TI - Understanding Neural Networks Through Deep Visualization
AU - Yosinski, Jason
AU - Clune, Jeff
AU - Nguyen, Anh
AU - Fuchs, Thomas
AU - Lipson, Hod
AB - Recent years have produced great advances in training large, deep neural networks (DNNs), including notable successes in training convolutional neural networks (convnets) to recognize natural images. However, our understanding of how these models work, especially what computations they perform at intermediate layers, has lagged behind. Progress in the field will be further accelerated by the development of better tools for visualizing and interpreting neural nets. We introduce two such tools here. The first is a tool that visualizes the activations produced on each layer of a trained convnet as it processes an image or video (e.g. a live webcam stream). We have found that looking at live activations that change in response to user input helps build valuable intuitions about how convnets work. The second tool enables visualizing features at each layer of a DNN via regularized optimization in image space. Because previous versions of this idea produced less recognizable images, here we introduce several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations. Both tools are open source and work on a pre-trained convnet with minimal setup.
C3 - ICML Deep Learning Workshop
DA - 2015///
PY - 2015
DP - Google Scholar
UR - http://arxiv.org/abs/1506.06579
Y2 - 2015/12/03/01:13:42
ER -

Recent years have produced great advances in training large, deep neural networks (DNNs), including notable successes in training convolutional neural networks (convnets) to recognize natural images. However, our understanding of how these models work, especially what computations they perform at intermediate layers, has lagged behind. Progress in the field will be further accelerated by the development of better tools for visualizing and interpreting neural nets. We introduce two such tools here. The first is a tool that visualizes the activations produced on each layer of a trained convnet as it processes an image or video (e.g. a live webcam stream). We have found that looking at live activations that change in response to user input helps build valuable intuitions about how convnets work. The second tool enables visualizing features at each layer of a DNN via regularized optimization in image space. Because previous versions of this idea produced less recognizable images, here we introduce several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations. Both tools are open source and work on a pre-trained convnet with minimal setup.

Enhanced Flyby Science with Onboard Computer Vision: Tracking and Surface Feature Detection at Small Bodies.

Thomas J. Fuchs, David R. Thompson, Brian D. Bue, Julie Castillo-Rogez, Steve A. Chien, Dero Gharibian and Kiri L. Wagstaff.

Earth and Space Science, vol. 2, 10, p. 417-34, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{fuchs_enhanced_2015,
    title = {Enhanced {Flyby} {Science} with {Onboard} {Computer} {Vision}: {Tracking} and {Surface} {Feature} {Detection} at {Small} {Bodies}},
    volume = {2},
    issn = {2333-5084},
    shorttitle = {Enhanced flyby science with onboard computer vision},
    url = {http://onlinelibrary.wiley.com/doi/10.1002/2014EA000042/abstract},
    doi = {10.1002/2014EA000042},
    abstract = {Spacecraft autonomy is crucial to increase the science return of optical remote sensing observations at distant primitive bodies. To date, most small bodies exploration has involved short timescale flybys that execute prescripted data collection sequences. Light time delay means that the spacecraft must operate completely autonomously without direct control from the ground, but in most cases the physical properties and morphologies of prospective targets are unknown before the flyby. Surface features of interest are highly localized, and successful observations must account for geometry and illumination constraints. Under these circumstances onboard computer vision can improve science yield by responding immediately to collected imagery. It can reacquire bad data or identify features of opportunity for additional targeted measurements. We present a comprehensive framework for onboard computer vision for flyby missions at small bodies. We introduce novel algorithms for target tracking, target segmentation, surface feature detection, and anomaly detection. The performance and generalization power are evaluated in detail using expert annotations on data sets from previous encounters with primitive bodies.},
    language = {en},
    number = {10},
    urldate = {2015-11-22},
    journal = {Earth and Space Science},
    author = {Fuchs, Thomas J. and Thompson, David R. and Bue, Brian D. and Castillo-Rogez, Julie and Chien, Steve A. and Gharibian, Dero and Wagstaff, Kiri L.},
    year = {2015},
    keywords = {0540 Image processing, 0555 Neural networks, fuzzy logic, machine learning, 6055 Surfaces, 6094 Instruments and techniques, 6205 Asteroids, asteroids, comets, computer vision, flyby, machine learning, small bodies},
    pages = {417--34},
}

Download Endnote/RIS citation

TY - JOUR
TI - Enhanced Flyby Science with Onboard Computer Vision: Tracking and Surface Feature Detection at Small Bodies
AU - Fuchs, Thomas J.
AU - Thompson, David R.
AU - Bue, Brian D.
AU - Castillo-Rogez, Julie
AU - Chien, Steve A.
AU - Gharibian, Dero
AU - Wagstaff, Kiri L.
T2 - Earth and Space Science
AB - Spacecraft autonomy is crucial to increase the science return of optical remote sensing observations at distant primitive bodies. To date, most small bodies exploration has involved short timescale flybys that execute prescripted data collection sequences. Light time delay means that the spacecraft must operate completely autonomously without direct control from the ground, but in most cases the physical properties and morphologies of prospective targets are unknown before the flyby. Surface features of interest are highly localized, and successful observations must account for geometry and illumination constraints. Under these circumstances onboard computer vision can improve science yield by responding immediately to collected imagery. It can reacquire bad data or identify features of opportunity for additional targeted measurements. We present a comprehensive framework for onboard computer vision for flyby missions at small bodies. We introduce novel algorithms for target tracking, target segmentation, surface feature detection, and anomaly detection. The performance and generalization power are evaluated in detail using expert annotations on data sets from previous encounters with primitive bodies.
DA - 2015///
PY - 2015
DO - 10.1002/2014EA000042
DP - Wiley Online Library
VL - 2
IS - 10
SP - 417
EP - 34
J2 - Earth and Space Science
LA - en
SN - 2333-5084
ST - Enhanced flyby science with onboard computer vision
UR - http://onlinelibrary.wiley.com/doi/10.1002/2014EA000042/abstract
Y2 - 2015/11/22/21:25:18
KW - 0540 Image processing
KW - 0555 Neural networks, fuzzy logic, machine learning
KW - 6055 Surfaces
KW - 6094 Instruments and techniques
KW - 6205 Asteroids
KW - asteroids
KW - comets
KW - computer vision
KW - flyby
KW - machine learning
KW - small bodies
ER -

Spacecraft autonomy is crucial to increase the science return of optical remote sensing observations at distant primitive bodies. To date, most small bodies exploration has involved short timescale flybys that execute prescripted data collection sequences. Light time delay means that the spacecraft must operate completely autonomously without direct control from the ground, but in most cases the physical properties and morphologies of prospective targets are unknown before the flyby. Surface features of interest are highly localized, and successful observations must account for geometry and illumination constraints. Under these circumstances onboard computer vision can improve science yield by responding immediately to collected imagery. It can reacquire bad data or identify features of opportunity for additional targeted measurements. We present a comprehensive framework for onboard computer vision for flyby missions at small bodies. We introduce novel algorithms for target tracking, target segmentation, surface feature detection, and anomaly detection. The performance and generalization power are evaluated in detail using expert annotations on data sets from previous encounters with primitive bodies.

Boosting Convolutional Features for Robust Object Proposals.

Nikolaos Karianakis, Thomas J. Fuchs and Stefano Soatto.

2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{karianakis_boosting_2015,
    title = {Boosting {Convolutional} {Features} for {Robust} {Object} {Proposals}},
    url = {http://arxiv.org/abs/1503.06350},
    author = {Karianakis, Nikolaos and Fuchs, Thomas J. and Soatto, Stefano},
    year = {2015},
}

Download Endnote/RIS citation

TY - JOUR
TI - Boosting Convolutional Features for Robust Object Proposals
AU - Karianakis, Nikolaos
AU - Fuchs, Thomas J.
AU - Soatto, Stefano
DA - 2015///
PY - 2015
UR - http://arxiv.org/abs/1503.06350
ER -

Machine Learning Approaches to Analyze Histological Images of Tissues from Radical Prostatectomies.

Arkadiusz Gertych, Nathan Ing, Zhaoxuan Ma, Thomas J. Fuchs, Sadri Salman, Sambit Mohanty, Sanica Bhele, Adriana Velásquez-Vacca, Mahul B. Amin and Beatrice S. Knudsen.

Computerized Medical Imaging and Graphics, vol. 46, Part 2, p. 197-208, Information Technologies in Biomedicine, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{gertych_machine_2015,
    series = {Information {Technologies} in {Biomedicine}},
    title = {Machine {Learning} {Approaches} to {Analyze} {Histological} {Images} of {Tissues} from {Radical} {Prostatectomies}},
    volume = {46, Part 2},
    issn = {0895-6111},
    url = {http://www.sciencedirect.com/science/article/pii/S0895611115001184},
    doi = {10.1016/j.compmedimag.2015.08.002},
    abstract = {Computerized evaluation of histological preparations of prostate tissues involves identification of tissue components such as stroma (ST), benign/normal epithelium (BN) and prostate cancer (PCa). Image classification approaches have been developed to identify and classify glandular regions in digital images of prostate tissues; however their success has been limited by difficulties in cellular segmentation and tissue heterogeneity. We hypothesized that utilizing image pixels to generate intensity histograms of hematoxylin (H) and eosin (E) stains deconvoluted from H\&E images numerically captures the architectural difference between glands and stroma. In addition, we postulated that joint histograms of local binary patterns and local variance (LBPxVAR) can be used as sensitive textural features to differentiate benign/normal tissue from cancer. Here we utilized a machine learning approach comprising of a support vector machine (SVM) followed by a random forest (RF) classifier to digitally stratify prostate tissue into ST, BN and PCa areas. Two pathologists manually annotated 210 images of low- and high-grade tumors from slides that were selected from 20 radical prostatectomies and digitized at high-resolution. The 210 images were split into the training (n = 19) and test (n = 191) sets. Local intensity histograms of H and E were used to train a SVM classifier to separate ST from epithelium (BN + PCa). The performance of SVM prediction was evaluated by measuring the accuracy of delineating epithelial areas. The Jaccard J = 59.5 ± 14.6 and Rand Ri = 62.0 ± 7.5 indices reported a significantly better prediction when compared to a reference method (Chen et al., Clinical Proteomics 2013, 10:18) based on the averaged values from the test set. To distinguish BN from PCa we trained a RF classifier with LBPxVAR and local intensity histograms and obtained separate performance values for BN and PCa: JBN = 35.2 ± 24.9, OBN = 49.6 ± 32, JPCa = 49.5 ± 18.5, OPCa = 72.7 ± 14.8 and Ri = 60.6 ± 7.6 in the test set. Our pixel-based classification does not rely on the detection of lumens, which is prone to errors and has limitations in high-grade cancers and has the potential to aid in clinical studies in which the quantification of tumor content is necessary to prognosticate the course of the disease. The image data set with ground truth annotation is available for public use to stimulate further research in this area.},
    urldate = {2016-01-03},
    journal = {Computerized Medical Imaging and Graphics},
    author = {Gertych, Arkadiusz and Ing, Nathan and Ma, Zhaoxuan and Fuchs, Thomas J. and Salman, Sadri and Mohanty, Sambit and Bhele, Sanica and Vel\'asquez-Vacca, Adriana and Amin, Mahul B. and Knudsen, Beatrice S.},
    year = {2015},
    keywords = {Image analysis, Prostate cancer, Tissue classification, Tissue quantification, machine learning},
    pages = {197--208},
}

Download Endnote/RIS citation

TY - JOUR
TI - Machine Learning Approaches to Analyze Histological Images of Tissues from Radical Prostatectomies
AU - Gertych, Arkadiusz
AU - Ing, Nathan
AU - Ma, Zhaoxuan
AU - Fuchs, Thomas J.
AU - Salman, Sadri
AU - Mohanty, Sambit
AU - Bhele, Sanica
AU - Velásquez-Vacca, Adriana
AU - Amin, Mahul B.
AU - Knudsen, Beatrice S.
T2 - Computerized Medical Imaging and Graphics
T3 - Information Technologies in Biomedicine
AB - Computerized evaluation of histological preparations of prostate tissues involves identification of tissue components such as stroma (ST), benign/normal epithelium (BN) and prostate cancer (PCa). Image classification approaches have been developed to identify and classify glandular regions in digital images of prostate tissues; however their success has been limited by difficulties in cellular segmentation and tissue heterogeneity. We hypothesized that utilizing image pixels to generate intensity histograms of hematoxylin (H) and eosin (E) stains deconvoluted from H&E images numerically captures the architectural difference between glands and stroma. In addition, we postulated that joint histograms of local binary patterns and local variance (LBPxVAR) can be used as sensitive textural features to differentiate benign/normal tissue from cancer. Here we utilized a machine learning approach comprising of a support vector machine (SVM) followed by a random forest (RF) classifier to digitally stratify prostate tissue into ST, BN and PCa areas. Two pathologists manually annotated 210 images of low- and high-grade tumors from slides that were selected from 20 radical prostatectomies and digitized at high-resolution. The 210 images were split into the training (n = 19) and test (n = 191) sets. Local intensity histograms of H and E were used to train a SVM classifier to separate ST from epithelium (BN + PCa). The performance of SVM prediction was evaluated by measuring the accuracy of delineating epithelial areas. The Jaccard J = 59.5 ± 14.6 and Rand Ri = 62.0 ± 7.5 indices reported a significantly better prediction when compared to a reference method (Chen et al., Clinical Proteomics 2013, 10:18) based on the averaged values from the test set. To distinguish BN from PCa we trained a RF classifier with LBPxVAR and local intensity histograms and obtained separate performance values for BN and PCa: JBN = 35.2 ± 24.9, OBN = 49.6 ± 32, JPCa = 49.5 ± 18.5, OPCa = 72.7 ± 14.8 and Ri = 60.6 ± 7.6 in the test set. Our pixel-based classification does not rely on the detection of lumens, which is prone to errors and has limitations in high-grade cancers and has the potential to aid in clinical studies in which the quantification of tumor content is necessary to prognosticate the course of the disease. The image data set with ground truth annotation is available for public use to stimulate further research in this area.
DA - 2015///
PY - 2015
DO - 10.1016/j.compmedimag.2015.08.002
DP - ScienceDirect
VL - 46, Part 2
SP - 197
EP - 208
J2 - Computerized Medical Imaging and Graphics
SN - 0895-6111
UR - http://www.sciencedirect.com/science/article/pii/S0895611115001184
Y2 - 2016/01/03/16:12:47
KW - Image analysis
KW - Prostate cancer
KW - Tissue classification
KW - Tissue quantification
KW - machine learning
ER -

Computerized evaluation of histological preparations of prostate tissues involves identification of tissue components such as stroma (ST), benign/normal epithelium (BN) and prostate cancer (PCa). Image classification approaches have been developed to identify and classify glandular regions in digital images of prostate tissues; however their success has been limited by difficulties in cellular segmentation and tissue heterogeneity. We hypothesized that utilizing image pixels to generate intensity histograms of hematoxylin (H) and eosin (E) stains deconvoluted from H&E images numerically captures the architectural difference between glands and stroma. In addition, we postulated that joint histograms of local binary patterns and local variance (LBPxVAR) can be used as sensitive textural features to differentiate benign/normal tissue from cancer. Here we utilized a machine learning approach comprising of a support vector machine (SVM) followed by a random forest (RF) classifier to digitally stratify prostate tissue into ST, BN and PCa areas. Two pathologists manually annotated 210 images of low- and high-grade tumors from slides that were selected from 20 radical prostatectomies and digitized at high-resolution. The 210 images were split into the training (n = 19) and test (n = 191) sets. Local intensity histograms of H and E were used to train a SVM classifier to separate ST from epithelium (BN + PCa). The performance of SVM prediction was evaluated by measuring the accuracy of delineating epithelial areas. The Jaccard J = 59.5 ± 14.6 and Rand Ri = 62.0 ± 7.5 indices reported a significantly better prediction when compared to a reference method (Chen et al., Clinical Proteomics 2013, 10:18) based on the averaged values from the test set. To distinguish BN from PCa we trained a RF classifier with LBPxVAR and local intensity histograms and obtained separate performance values for BN and PCa: JBN = 35.2 ± 24.9, OBN = 49.6 ± 32, JPCa = 49.5 ± 18.5, OPCa = 72.7 ± 14.8 and Ri = 60.6 ± 7.6 in the test set. Our pixel-based classification does not rely on the detection of lumens, which is prone to errors and has limitations in high-grade cancers and has the potential to aid in clinical studies in which the quantification of tumor content is necessary to prognosticate the course of the disease. The image data set with ground truth annotation is available for public use to stimulate further research in this area.

Risk-aware Planetary Rover Operation: Autonomous Terrain Classification and Path Planning.

Masahiro Ono, Thomas J. Fuchs, Amanda Steffy, Mark Maimone and Jeng Yen.

Proceedings of the 36th IEEE Aerospace Conference, p. 1–10, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{ono_risk-aware_2015,
    title = {Risk-aware {Planetary} {Rover} {Operation}: {Autonomous} {Terrain} {Classification} and {Path} {Planning}},
    url = {http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=7119022&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D7119022},
    doi = {10.1109/AERO.2015.7119022},
    abstract = {Identifying and avoiding terrain hazards (e.g., soft soil and pointy embedded rocks) are crucial for the safety of planetary rovers. This paper presents a newly developed ground-based Mars rover operation tool that mitigates risks from terrain by automatically identifying hazards on the terrain, evaluating their risks, and suggesting operators safe paths options that avoids potential risks while achieving specified goals. The tool will bring benefits to rover operations by reducing operation cost, by reducing cognitive load of rover operators, by preventing human errors, and most importantly, by significantly reducing the risk of the loss of rovers. The risk-aware rover operation tool is built upon two technologies. The first technology is a machine learning-based terrain classification that is capable of identifying potential hazards, such as pointy rocks and soft terrains, from images. The second technology is a risk-aware path planner based on rapidly-exploring random graph (RRG) and the A* search algorithms, which is capable of avoiding hazards identified by the terrain classifier with explicitly considering wheel placement. We demonstrate the integrated capability of the proposed risk-aware rover operation tool by using the images taken by the Curiosity rover.},
    booktitle = {Proceedings of the 36th {IEEE} {Aerospace} {Conference}},
    author = {Ono, Masahiro and Fuchs, Thomas J. and Steffy, Amanda and Maimone, Mark and Yen, Jeng},
    year = {2015},
    pages = {1--10},
}

Download Endnote/RIS citation

TY - CONF
TI - Risk-aware Planetary Rover Operation: Autonomous Terrain Classification and Path Planning
AU - Ono, Masahiro
AU - Fuchs, Thomas J.
AU - Steffy, Amanda
AU - Maimone, Mark
AU - Yen, Jeng
AB - Identifying and avoiding terrain hazards (e.g., soft soil and pointy embedded rocks) are crucial for the safety of planetary rovers. This paper presents a newly developed ground-based Mars rover operation tool that mitigates risks from terrain by automatically identifying hazards on the terrain, evaluating their risks, and suggesting operators safe paths options that avoids potential risks while achieving specified goals. The tool will bring benefits to rover operations by reducing operation cost, by reducing cognitive load of rover operators, by preventing human errors, and most importantly, by significantly reducing the risk of the loss of rovers. The risk-aware rover operation tool is built upon two technologies. The first technology is a machine learning-based terrain classification that is capable of identifying potential hazards, such as pointy rocks and soft terrains, from images. The second technology is a risk-aware path planner based on rapidly-exploring random graph (RRG) and the A* search algorithms, which is capable of avoiding hazards identified by the terrain classifier with explicitly considering wheel placement. We demonstrate the integrated capability of the proposed risk-aware rover operation tool by using the images taken by the Curiosity rover.
C3 - Proceedings of the 36th IEEE Aerospace Conference
DA - 2015///
PY - 2015
DO - 10.1109/AERO.2015.7119022
SP - 1
EP - 10
UR - http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=7119022&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D7119022
ER -

Identifying and avoiding terrain hazards (e.g., soft soil and pointy embedded rocks) are crucial for the safety of planetary rovers. This paper presents a newly developed ground-based Mars rover operation tool that mitigates risks from terrain by automatically identifying hazards on the terrain, evaluating their risks, and suggesting operators safe paths options that avoids potential risks while achieving specified goals. The tool will bring benefits to rover operations by reducing operation cost, by reducing cognitive load of rover operators, by preventing human errors, and most importantly, by significantly reducing the risk of the loss of rovers. The risk-aware rover operation tool is built upon two technologies. The first technology is a machine learning-based terrain classification that is capable of identifying potential hazards, such as pointy rocks and soft terrains, from images. The second technology is a risk-aware path planner based on rapidly-exploring random graph (RRG) and the A* search algorithms, which is capable of avoiding hazards identified by the terrain classifier with explicitly considering wheel placement. We demonstrate the integrated capability of the proposed risk-aware rover operation tool by using the images taken by the Curiosity rover.

Robot-Centric Activity Prediction from First-Person Videos: What Will They Do to Me?

Michael S. Ryoo, Thomas J. Fuchs, Lu Xia, J. K. Aggarwal and Larry H. Matthies.

Proceedings of the 10th ACM/IEEE International Conference on Human-Robot Interaction, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{ryoo_robot-centric_2015,
    title = {Robot-{Centric} {Activity} {Prediction} from {First}-{Person} {Videos}: {What} {Will} {They} {Do} to {Me}?},
    url = {http://michaelryoo.com/papers/hri2015_ryoo.pdf},
    booktitle = {Proceedings of the 10th {ACM}/{IEEE} {International} {Conference} on {Human}-{Robot} {Interaction}},
    author = {Ryoo, Michael S. and Fuchs, Thomas J. and Xia, Lu and Aggarwal, J. K. and Matthies, Larry H.},
    year = {2015},
}

Download Endnote/RIS citation

TY - CONF
TI - Robot-Centric Activity Prediction from First-Person Videos: What Will They Do to Me?
AU - Ryoo, Michael S.
AU - Fuchs, Thomas J.
AU - Xia, Lu
AU - Aggarwal, J. K.
AU - Matthies, Larry H.
C3 - Proceedings of the 10th ACM/IEEE International Conference on Human-Robot Interaction
DA - 2015///
PY - 2015
UR - http://michaelryoo.com/papers/hri2015_ryoo.pdf
ER -

Crohn's Disease Segmentation from MRI Using Learned Image Priors.

D. Mahapatra, P. J. Schüffler, F. M. Vos and J. M. Buhmann.

Proceedings IEEE ISBI 2015, p. 625-628, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{mahapatra_crohns_2015,
    title = {Crohn's {Disease} {Segmentation} from {MRI} {Using} {Learned} {Image} {Priors}},
    url = {http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=7163951},
    abstract = {We use a Field of Experts (FoE) model to segment abdominal regions from MRI affected with Crohns Disease (CD). FoE learns a prior model of diseased and normal bowel, and background non-bowel tissues from manually annotated training images. Unlike current approaches, FoE does not rely on hand designed features but learns the most discriminative features (in the form of filters) for different classes. FoE filter responses are integrated into a Random forest (RF) model that outputs probability maps for the test image and finally segments the diseased region. Experimental results show our method achieves significantly better performance than existing methods.},
    journal = {Proceedings IEEE ISBI 2015},
    author = {Mahapatra, D. and Sch\"uffler, P. J. and Vos, F. M. and Buhmann, J. M.},
    year = {2015},
    pages = {625--628},
}

Download Endnote/RIS citation

TY - JOUR
TI - Crohn's Disease Segmentation from MRI Using Learned Image Priors
AU - Mahapatra, D.
AU - Schüffler, P. J.
AU - Vos, F. M.
AU - Buhmann, J. M.
T2 - Proceedings IEEE ISBI 2015
AB - We use a Field of Experts (FoE) model to segment abdominal regions from MRI affected with Crohns Disease (CD). FoE learns a prior model of diseased and normal bowel, and background non-bowel tissues from manually annotated training images. Unlike current approaches, FoE does not rely on hand designed features but learns the most discriminative features (in the form of filters) for different classes. FoE filter responses are integrated into a Random forest (RF) model that outputs probability maps for the test image and finally segments the diseased region. Experimental results show our method achieves significantly better performance than existing methods.
DA - 2015///
PY - 2015
SP - 625
EP - 628
UR - http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=7163951
ER -

We use a Field of Experts (FoE) model to segment abdominal regions from MRI affected with Crohns Disease (CD). FoE learns a prior model of diseased and normal bowel, and background non-bowel tissues from manually annotated training images. Unlike current approaches, FoE does not rely on hand designed features but learns the most discriminative features (in the form of filters) for different classes. FoE filter responses are integrated into a Random forest (RF) model that outputs probability maps for the test image and finally segments the diseased region. Experimental results show our method achieves significantly better performance than existing methods.

Autonomous Real-time Detection of Plumes and Jets from Moons and Comets.

Kiri L. Wagstaff, David R. Thompson, Brian D. Bue and Thomas J. Fuchs.

ApJ, vol. 794, 1, p. 43, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{wagstaff_autonomous_2014,
    title = {Autonomous {Real}-time {Detection} of {Plumes} and {Jets} from {Moons} and {Comets}},
    volume = {794},
    issn = {0004-637X},
    url = {http://stacks.iop.org/0004-637X/794/i=1/a=43},
    doi = {10.1088/0004-637X/794/1/43},
    abstract = {Dynamic activity on the surface of distant moons, asteroids, and comets can manifest as jets or plumes. These phenomena provide information about the interior of the bodies and the forces (gravitation, radiation, thermal) they experience. Fast detection and follow-up study is imperative since the phenomena may be time-varying and because the observing window may be limited (e.g., during a flyby). We have developed an advanced method for real-time detection of plumes and jets using onboard analysis of the data as it is collected. In contrast to prior work, our technique is not restricted to plume detection from spherical bodies, making it relevant for irregularly shaped bodies such as comets. Further, our study analyzes raw data, the form in which it is available on board the spacecraft, rather than fully processed image products. In summary, we contribute a vital assessment of a technique that can be used on board tomorrow's deep space missions to detect, and respond quickly to, new occurrences of plumes and jets.},
    language = {en},
    number = {1},
    urldate = {2016-01-05},
    journal = {The Astrophysical Journal},
    author = {Wagstaff, Kiri L. and Thompson, David R. and Bue, Brian D. and Fuchs, Thomas J.},
    year = {2014},
    pages = {43},
}

Download Endnote/RIS citation

TY - JOUR
TI - Autonomous Real-time Detection of Plumes and Jets from Moons and Comets
AU - Wagstaff, Kiri L.
AU - Thompson, David R.
AU - Bue, Brian D.
AU - Fuchs, Thomas J.
T2 - The Astrophysical Journal
AB - Dynamic activity on the surface of distant moons, asteroids, and comets can manifest as jets or plumes. These phenomena provide information about the interior of the bodies and the forces (gravitation, radiation, thermal) they experience. Fast detection and follow-up study is imperative since the phenomena may be time-varying and because the observing window may be limited (e.g., during a flyby). We have developed an advanced method for real-time detection of plumes and jets using onboard analysis of the data as it is collected. In contrast to prior work, our technique is not restricted to plume detection from spherical bodies, making it relevant for irregularly shaped bodies such as comets. Further, our study analyzes raw data, the form in which it is available on board the spacecraft, rather than fully processed image products. In summary, we contribute a vital assessment of a technique that can be used on board tomorrow's deep space missions to detect, and respond quickly to, new occurrences of plumes and jets.
DA - 2014///
PY - 2014
DO - 10.1088/0004-637X/794/1/43
DP - Institute of Physics
VL - 794
IS - 1
SP - 43
J2 - ApJ
LA - en
SN - 0004-637X
UR - http://stacks.iop.org/0004-637X/794/i=1/a=43
Y2 - 2016/01/05/03:51:25
ER -

Dynamic activity on the surface of distant moons, asteroids, and comets can manifest as jets or plumes. These phenomena provide information about the interior of the bodies and the forces (gravitation, radiation, thermal) they experience. Fast detection and follow-up study is imperative since the phenomena may be time-varying and because the observing window may be limited (e.g., during a flyby). We have developed an advanced method for real-time detection of plumes and jets using onboard analysis of the data as it is collected. In contrast to prior work, our technique is not restricted to plume detection from spherical bodies, making it relevant for irregularly shaped bodies such as comets. Further, our study analyzes raw data, the form in which it is available on board the spacecraft, rather than fully processed image products. In summary, we contribute a vital assessment of a technique that can be used on board tomorrow's deep space missions to detect, and respond quickly to, new occurrences of plumes and jets.

Sparse Meta-Gaussian Information Bottleneck.

Melanie Rey, Thomas J. Fuchs and Volker Roth.

Proceedings of the 31st International Conference on Machine Learning, ICML, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{rey_sparse_2014,
    series = {{ICML}},
    title = {Sparse {Meta}-{Gaussian} {Information} {Bottleneck}},
    url = {http://jmlr.org/proceedings/papers/v32/rey14.pdf},
    booktitle = {Proceedings of the 31st {International} {Conference} on {Machine} {Learning}},
    author = {Rey, Melanie and Fuchs, Thomas J. and Roth, Volker},
    year = {2014},
}

Download Endnote/RIS citation

TY - CONF
TI - Sparse Meta-Gaussian Information Bottleneck
AU - Rey, Melanie
AU - Fuchs, Thomas J.
AU - Roth, Volker
T3 - ICML
C3 - Proceedings of the 31st International Conference on Machine Learning
DA - 2014///
PY - 2014
UR - http://jmlr.org/proceedings/papers/v32/rey14.pdf
ER -

KPNA2 Is Overexpressed in Human and Mouse Endometrial Cancers and Promotes Cellular Proliferation.

Kristian Ikenberg, Nadejda Valtcheva, Simone Brandt, Qing Zhong, Christine E. Wong, Aurelia Noske, Markus Rechsteiner, Jan H. Rueschoff, Rosemarie Caduff, Athanassios Dellas, Ellen Obermann, Daniel Fink, Thomas J. Fuchs, Wilhelm Krek, Holger Moch, Ian J. Frew and Peter J. Wild.

The Journal of Pathology, vol. 234, 2, p. 239–252, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{ikenberg_kpna2_2014,
    title = {{KPNA2} {Is} {Overexpressed} in {Human} and {Mouse} {Endometrial} {Cancers} and {Promotes} {Cellular} {Proliferation}},
    volume = {234},
    issn = {1096-9896},
    url = {http://dx.doi.org/10.1002/path.4390},
    doi = {10.1002/path.4390},
    number = {2},
    journal = {The Journal of Pathology},
    author = {Ikenberg, Kristian and Valtcheva, Nadejda and Brandt, Simone and Zhong, Qing and Wong, Christine E. and Noske, Aurelia and Rechsteiner, Markus and Rueschoff, Jan H. and Caduff, Rosemarie and Dellas, Athanassios and Obermann, Ellen and Fink, Daniel and Fuchs, Thomas J. and Krek, Wilhelm and Moch, Holger and Frew, Ian J. and Wild, Peter J.},
    year = {2014},
    keywords = {EMT, KPNA2, Snail, biomarker, endometrial cancer, importin},
    pages = {239--252},
}

Download Endnote/RIS citation

TY - JOUR
TI - KPNA2 Is Overexpressed in Human and Mouse Endometrial Cancers and Promotes Cellular Proliferation
AU - Ikenberg, Kristian
AU - Valtcheva, Nadejda
AU - Brandt, Simone
AU - Zhong, Qing
AU - Wong, Christine E.
AU - Noske, Aurelia
AU - Rechsteiner, Markus
AU - Rueschoff, Jan H.
AU - Caduff, Rosemarie
AU - Dellas, Athanassios
AU - Obermann, Ellen
AU - Fink, Daniel
AU - Fuchs, Thomas J.
AU - Krek, Wilhelm
AU - Moch, Holger
AU - Frew, Ian J.
AU - Wild, Peter J.
T2 - The Journal of Pathology
DA - 2014///
PY - 2014
DO - 10.1002/path.4390
VL - 234
IS - 2
SP - 239
EP - 252
SN - 1096-9896
UR - http://dx.doi.org/10.1002/path.4390
KW - EMT
KW - KPNA2
KW - Snail
KW - biomarker
KW - endometrial cancer
KW - importin
ER -

Highly Multiplexed Imaging of Tumor Tissues with Subcellular Resolution by Mass Cytometry.

C. Giesen, H. A. Wang, D. Schapiro, N. Zivanovic, A. Jacobs, B. Hattendorf, P. J. Schüffler, D. Grolimund, J. M. Buhmann, S. Brandt, Z. Varga, P. J. Wild, D. Gunther and B. Bodenmiller.

Nature methods, vol. 11, p. 417-22, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{giesen_highly_2014,
    title = {Highly {Multiplexed} {Imaging} of {Tumor} {Tissues} with {Subcellular} {Resolution} by {Mass} {Cytometry}},
    volume = {11},
    issn = {1548-7105 (Electronic) 1548-7091 (Linking)},
    url = {http://www.nature.com/nmeth/journal/v11/n4/full/nmeth.2869.html},
    doi = {10.1038/nmeth.2869},
    abstract = {Mass cytometry enables high-dimensional, single-cell analysis of cell type and state. In mass cytometry, rare earth metals are used as reporters on antibodies. Analysis of metal abundances using the mass cytometer allows determination of marker expression in individual cells. Mass cytometry has previously been applied only to cell suspensions. To gain spatial information, we have coupled immunohistochemical and immunocytochemical methods with high-resolution laser ablation to CyTOF mass cytometry. This approach enables the simultaneous imaging of 32 proteins and protein modifications at subcellular resolution; with the availability of additional isotopes, measurement of over 100 markers will be possible. We applied imaging mass cytometry to human breast cancer samples, allowing delineation of cell subpopulations and cell-cell interactions and highlighting tumor heterogeneity. Imaging mass cytometry complements existing imaging approaches. It will enable basic studies of tissue heterogeneity and function and support the transition of medicine toward individualized molecularly targeted diagnosis and therapies.},
    journal = {Nat Methods},
    author = {Giesen, C. and Wang, H. A. and Schapiro, D. and Zivanovic, N. and Jacobs, A. and Hattendorf, B. and Sch\"uffler, P. J. and Grolimund, D. and Buhmann, J. M. and Brandt, S. and Varga, Z. and Wild, P. J. and Gunther, D. and Bodenmiller, B.},
    year = {2014},
    pages = {417--22},
}

Download Endnote/RIS citation

TY - JOUR
TI - Highly Multiplexed Imaging of Tumor Tissues with Subcellular Resolution by Mass Cytometry
AU - Giesen, C.
AU - Wang, H. A.
AU - Schapiro, D.
AU - Zivanovic, N.
AU - Jacobs, A.
AU - Hattendorf, B.
AU - Schüffler, P. J.
AU - Grolimund, D.
AU - Buhmann, J. M.
AU - Brandt, S.
AU - Varga, Z.
AU - Wild, P. J.
AU - Gunther, D.
AU - Bodenmiller, B.
T2 - Nat Methods
AB - Mass cytometry enables high-dimensional, single-cell analysis of cell type and state. In mass cytometry, rare earth metals are used as reporters on antibodies. Analysis of metal abundances using the mass cytometer allows determination of marker expression in individual cells. Mass cytometry has previously been applied only to cell suspensions. To gain spatial information, we have coupled immunohistochemical and immunocytochemical methods with high-resolution laser ablation to CyTOF mass cytometry. This approach enables the simultaneous imaging of 32 proteins and protein modifications at subcellular resolution; with the availability of additional isotopes, measurement of over 100 markers will be possible. We applied imaging mass cytometry to human breast cancer samples, allowing delineation of cell subpopulations and cell-cell interactions and highlighting tumor heterogeneity. Imaging mass cytometry complements existing imaging approaches. It will enable basic studies of tissue heterogeneity and function and support the transition of medicine toward individualized molecularly targeted diagnosis and therapies.
DA - 2014///
PY - 2014
DO - 10.1038/nmeth.2869
VL - 11
SP - 417
EP - 22
J2 - Nature methods
SN - 1548-7105 (Electronic) 1548-7091 (Linking)
UR - http://www.nature.com/nmeth/journal/v11/n4/full/nmeth.2869.html
ER -

Mass cytometry enables high-dimensional, single-cell analysis of cell type and state. In mass cytometry, rare earth metals are used as reporters on antibodies. Analysis of metal abundances using the mass cytometer allows determination of marker expression in individual cells. Mass cytometry has previously been applied only to cell suspensions. To gain spatial information, we have coupled immunohistochemical and immunocytochemical methods with high-resolution laser ablation to CyTOF mass cytometry. This approach enables the simultaneous imaging of 32 proteins and protein modifications at subcellular resolution; with the availability of additional isotopes, measurement of over 100 markers will be possible. We applied imaging mass cytometry to human breast cancer samples, allowing delineation of cell subpopulations and cell-cell interactions and highlighting tumor heterogeneity. Imaging mass cytometry complements existing imaging approaches. It will enable basic studies of tissue heterogeneity and function and support the transition of medicine toward individualized molecularly targeted diagnosis and therapies.

Automated Real-Time Classification and Decision Making in Massive Data Streams from Synoptic Sky Surveys.

S.G. Djorgovski, A. Mahabal, C. Donalek, M. Graham, A. Drake, M. Turmon and T.J. Fuchs.

2014 IEEE 10th International Conference on e-Science (e-Science), 2014 IEEE 10th International Conference on e-Science (e-Science), vol. 1, p. 204-211, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{djorgovski_automated_2014,
    title = {Automated {Real}-{Time} {Classification} and {Decision} {Making} in {Massive} {Data} {Streams} from {Synoptic} {Sky} {Surveys}},
    volume = {1},
    url = {http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6972266&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6972266},
    doi = {10.1109/eScience.2014.7},
    abstract = {The nature of scientific and technological data collection is evolving rapidly: data volumes and rates grow exponentially, with increasing complexity and information content, and there has been a transition from static data sets to data streams that must be analyzed in real time. Interesting or anomalous phenomena must be quickly characterized and followed up with additional measurements via optimal deployment of limited assets. Modern astronomy presents a variety of such phenomena in the form of transient events in digital synoptic sky surveys, including cosmic explosions (supernovae, gamma ray bursts), relativistic phenomena (black hole formation, jets), potentially hazardous asteroids, etc. We have been developing a set of machine learning tools to detect, classify and plan a response to transient events for astronomy applications, using the Catalina Real-time Transient Survey (CRTS) as a scientific and methodological testbed. The ability to respond rapidly to the potentially most interesting events is a key bottleneck that limits the scientific returns from the current and anticipated synoptic sky surveys. Similar challenge arise in other contexts, from environmental monitoring using sensor networks to autonomous spacecraft systems. Given the exponential growth of data rates, and the time-critical response, we need a fully automated and robust approach. We describe the results obtained to date, and the possible future developments.},
    booktitle = {2014 {IEEE} 10th {International} {Conference} on e-{Science} (e-{Science})},
    author = {Djorgovski, S.G. and Mahabal, A. and Donalek, C. and Graham, M. and Drake, A. and Turmon, M. and Fuchs, T.J.},
    year = {2014},
    keywords = {Astronomy, Automated decision making, Bayesian methods, CRTS, Catalina Real-time Transient Survey, Cathode ray tubes, Data analysis, Extraterrestrial measurements, Massive data streams, Pollution measurement, Real-time systems, Sky surveys, Time measurement, Transient analysis, astronomical surveys, astronomy applications, astronomy computing, automated real-time classification, automated real-time decision making, black hole formation, classification, cosmic explosions, decision making, digital synoptic sky surveys, gamma ray bursts, jets, learning (artificial intelligence), machine learning, machine learning tools, pattern classification, potentially hazardous asteroids, relativistic phenomena, scientific data collection, supernovae, technological data collection},
    pages = {204--211},
}

Download Endnote/RIS citation

TY - CONF
TI - Automated Real-Time Classification and Decision Making in Massive Data Streams from Synoptic Sky Surveys
AU - Djorgovski, S.G.
AU - Mahabal, A.
AU - Donalek, C.
AU - Graham, M.
AU - Drake, A.
AU - Turmon, M.
AU - Fuchs, T.J.
T2 - 2014 IEEE 10th International Conference on e-Science (e-Science)
AB - The nature of scientific and technological data collection is evolving rapidly: data volumes and rates grow exponentially, with increasing complexity and information content, and there has been a transition from static data sets to data streams that must be analyzed in real time. Interesting or anomalous phenomena must be quickly characterized and followed up with additional measurements via optimal deployment of limited assets. Modern astronomy presents a variety of such phenomena in the form of transient events in digital synoptic sky surveys, including cosmic explosions (supernovae, gamma ray bursts), relativistic phenomena (black hole formation, jets), potentially hazardous asteroids, etc. We have been developing a set of machine learning tools to detect, classify and plan a response to transient events for astronomy applications, using the Catalina Real-time Transient Survey (CRTS) as a scientific and methodological testbed. The ability to respond rapidly to the potentially most interesting events is a key bottleneck that limits the scientific returns from the current and anticipated synoptic sky surveys. Similar challenge arise in other contexts, from environmental monitoring using sensor networks to autonomous spacecraft systems. Given the exponential growth of data rates, and the time-critical response, we need a fully automated and robust approach. We describe the results obtained to date, and the possible future developments.
C3 - 2014 IEEE 10th International Conference on e-Science (e-Science)
DA - 2014///
PY - 2014
DO - 10.1109/eScience.2014.7
DP - IEEE Xplore
VL - 1
SP - 204
EP - 211
UR - http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6972266&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6972266
KW - Astronomy
KW - Automated decision making
KW - Bayesian methods
KW - CRTS
KW - Catalina Real-time Transient Survey
KW - Cathode ray tubes
KW - Data analysis
KW - Extraterrestrial measurements
KW - Massive data streams
KW - Pollution measurement
KW - Real-time systems
KW - Sky surveys
KW - Time measurement
KW - Transient analysis
KW - astronomical surveys
KW - astronomy applications
KW - astronomy computing
KW - automated real-time classification
KW - automated real-time decision making
KW - black hole formation
KW - classification
KW - cosmic explosions
KW - decision making
KW - digital synoptic sky surveys
KW - gamma ray bursts
KW - jets
KW - learning (artificial intelligence)
KW - machine learning
KW - machine learning tools
KW - pattern classification
KW - potentially hazardous asteroids
KW - relativistic phenomena
KW - scientific data collection
KW - supernovae
KW - technological data collection
ER -

Early Recognition of Human Activities from First-Person Videos Using Onset Representations.

M. S. Ryoo, Thomas J. Fuchs, Lu Xia, J. K. Aggarwal and Larry Matthies.

arXiv:1406.5309 [cs], 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{ryoo_early_2014,
    title = {Early {Recognition} of {Human} {Activities} from {First}-{Person} {Videos} {Using} {Onset} {Representations}},
    url = {http://arxiv.org/abs/1406.5309},
    abstract = {In this paper, we propose a methodology for early recognition of human activities from videos taken with a first-person viewpoint. Early recognition, which is also known as activity prediction, is an ability to infer an ongoing activity at its early stage. We present an algorithm to perform recognition of activities targeted at the camera from streaming videos, making the system to predict intended activities of the interacting person and avoid harmful events before they actually happen. We introduce the novel concept of 'onset' that efficiently summarizes pre-activity observations, and design an approach to consider event history in addition to ongoing video observation for early first-person recognition of activities. We propose to represent onset using cascade histograms of time series gradients, and we describe a novel algorithmic setup to take advantage of onset for early recognition of activities. The experimental results clearly illustrate that the proposed concept of onset enables better/earlier recognition of human activities from first-person videos.},
    urldate = {2015-12-23},
    journal = {arXiv:1406.5309 [cs]},
    author = {Ryoo, M. S. and Fuchs, Thomas J. and Xia, Lu and Aggarwal, J. K. and Matthies, Larry},
    year = {2014},
    keywords = {Computer Science - Computer Vision and Pattern Recognition},
}

Download Endnote/RIS citation

TY - JOUR
TI - Early Recognition of Human Activities from First-Person Videos Using Onset Representations
AU - Ryoo, M. S.
AU - Fuchs, Thomas J.
AU - Xia, Lu
AU - Aggarwal, J. K.
AU - Matthies, Larry
T2 - arXiv:1406.5309 [cs]
AB - In this paper, we propose a methodology for early recognition of human activities from videos taken with a first-person viewpoint. Early recognition, which is also known as activity prediction, is an ability to infer an ongoing activity at its early stage. We present an algorithm to perform recognition of activities targeted at the camera from streaming videos, making the system to predict intended activities of the interacting person and avoid harmful events before they actually happen. We introduce the novel concept of 'onset' that efficiently summarizes pre-activity observations, and design an approach to consider event history in addition to ongoing video observation for early first-person recognition of activities. We propose to represent onset using cascade histograms of time series gradients, and we describe a novel algorithmic setup to take advantage of onset for early recognition of activities. The experimental results clearly illustrate that the proposed concept of onset enables better/earlier recognition of human activities from first-person videos.
DA - 2014///
PY - 2014
DP - arXiv.org
UR - http://arxiv.org/abs/1406.5309
Y2 - 2015/12/23/16:01:25
KW - Computer Science - Computer Vision and Pattern Recognition
ER -

In this paper, we propose a methodology for early recognition of human activities from videos taken with a first-person viewpoint. Early recognition, which is also known as activity prediction, is an ability to infer an ongoing activity at its early stage. We present an algorithm to perform recognition of activities targeted at the camera from streaming videos, making the system to predict intended activities of the interacting person and avoid harmful events before they actually happen. We introduce the novel concept of 'onset' that efficiently summarizes pre-activity observations, and design an approach to consider event history in addition to ongoing video observation for early first-person recognition of activities. We propose to represent onset using cascade histograms of time series gradients, and we describe a novel algorithmic setup to take advantage of onset for early recognition of activities. The experimental results clearly illustrate that the proposed concept of onset enables better/earlier recognition of human activities from first-person videos.

Computer Aided Crohn's Disease Severity Assessment in MRI.

Peter J. Schüffler, Dwarikanath Mahapatra, Franciscus M. Vos and Joachim M. Buhmann.

VIGOR++ Workshop 2014 - Showcase of Research Outcomes and Future Outlook, 2014

Best Poster Award

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@misc{schuffler_computer_2014,
    address = {London},
    type = {Poster},
    title = {Computer {Aided} {Crohn}'s {Disease} {Severity} {Assessment} in {MRI}},
    url = {https://www.researchgate.net/publication/262198303_Computer_Aided_Crohns_Disease_Severity_Assessment_in_MRI},
    author = {Sch\"uffler, Peter J. and Mahapatra, Dwarikanath and Vos, Franciscus M. and Buhmann, Joachim M.},
    year = {2014},
}

Download Endnote/RIS citation

TY - SLIDE
TI - Computer Aided Crohn's Disease Severity Assessment in MRI
T2 - VIGOR++ Workshop 2014 - Showcase of Research Outcomes and Future Outlook
A2 - Schüffler, Peter J.
A2 - Mahapatra, Dwarikanath
A2 - Vos, Franciscus M.
A2 - Buhmann, Joachim M.
CY - London
DA - 2014///
PY - 2014
M3 - Poster
UR - https://www.researchgate.net/publication/262198303_Computer_Aided_Crohns_Disease_Severity_Assessment_in_MRI
ER -

Semi-Automatic Crohn's Disease Severity Estimation on MR Imaging.

P. J. Schüffler, D. Mahapatra, R. E. Naziroglu, Z. Li, C. A. J. Puylaert, R. Andriantsimiavona, F. M. Vos, D. A. Pendsé, C. Yung Nio, J. Stoker, S. A. Taylor and J. M. Buhmann.

6th MICCAI Workshop on Abdominal Imaging - Computational and Clinical Applications, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schuffler_semi-automatic_2014,
    title = {Semi-{Automatic} {Crohn}'s {Disease} {Severity} {Estimation} on {MR} {Imaging}},
    url = {http://link.springer.com/chapter/10.1007%2F978-3-319-13692-9_12},
    abstract = {Crohn’s disease (CD) is a chronic inflammatory bowel disease which can be visualized by magnetic resonance imaging (MRI). For CD grading, several non-invasive MRI based severity scores are known, most prominent the MaRIA and AIS. As these scores rely on manual MRI readings for individual bowel segments by trained radiologists, automated MRI assessment has been more and more focused in recent research. We show on a dataset of 27 CD patients that semi-automatically measured bowel wall thickness (ABWT) and dynamic contrast enhancement (DCE) completely outperform manual scorings: the segmental correlation to the Crohn’s Disease Endoscopic Index of Severity (CDEIS) of ABWT and DCE is significantly higher (r = .78) than that of MaRIA (r = .45) or AIS (r = .51). Also on a per-patient basis, the models with ABWT and DCE show significantly higher correlation (r = .69) to global CDEIS than MaRIA (r = .46).},
    author = {Sch\"uffler, P. J. and Mahapatra, D. and Naziroglu, R. E. and Li, Z. and Puylaert, C. A. J. and Andriantsimiavona, R. and Vos, F. M. and Pends\'e, D. A. and Nio, C. Yung and Stoker, J. and Taylor, S. A. and Buhmann, J. M.},
    year = {2014},
}

Download Endnote/RIS citation

TY - CONF
TI - Semi-Automatic Crohn's Disease Severity Estimation on MR Imaging
AU - Schüffler, P. J.
AU - Mahapatra, D.
AU - Naziroglu, R. E.
AU - Li, Z.
AU - Puylaert, C. A. J.
AU - Andriantsimiavona, R.
AU - Vos, F. M.
AU - Pendsé, D. A.
AU - Nio, C. Yung
AU - Stoker, J.
AU - Taylor, S. A.
AU - Buhmann, J. M.
T2 - 6th MICCAI Workshop on Abdominal Imaging - Computational and Clinical Applications
AB - Crohn’s disease (CD) is a chronic inflammatory bowel disease which can be visualized by magnetic resonance imaging (MRI). For CD grading, several non-invasive MRI based severity scores are known, most prominent the MaRIA and AIS. As these scores rely on manual MRI readings for individual bowel segments by trained radiologists, automated MRI assessment has been more and more focused in recent research. We show on a dataset of 27 CD patients that semi-automatically measured bowel wall thickness (ABWT) and dynamic contrast enhancement (DCE) completely outperform manual scorings: the segmental correlation to the Crohn’s Disease Endoscopic Index of Severity (CDEIS) of ABWT and DCE is significantly higher (r = .78) than that of MaRIA (r = .45) or AIS (r = .51). Also on a per-patient basis, the models with ABWT and DCE show significantly higher correlation (r = .69) to global CDEIS than MaRIA (r = .46).
DA - 2014///
PY - 2014
UR - http://link.springer.com/chapter/10.1007%2F978-3-319-13692-9_12
ER -

Crohn’s disease (CD) is a chronic inflammatory bowel disease which can be visualized by magnetic resonance imaging (MRI). For CD grading, several non-invasive MRI based severity scores are known, most prominent the MaRIA and AIS. As these scores rely on manual MRI readings for individual bowel segments by trained radiologists, automated MRI assessment has been more and more focused in recent research. We show on a dataset of 27 CD patients that semi-automatically measured bowel wall thickness (ABWT) and dynamic contrast enhancement (DCE) completely outperform manual scorings: the segmental correlation to the Crohn’s Disease Endoscopic Index of Severity (CDEIS) of ABWT and DCE is significantly higher (r = .78) than that of MaRIA (r = .45) or AIS (r = .51). Also on a per-patient basis, the models with ABWT and DCE show significantly higher correlation (r = .69) to global CDEIS than MaRIA (r = .46).

Autonomous Onboard Surface Feature Detection for Flyby Missions.

Thomas J. Fuchs, Brian D. Bue, Julie Castillo-Rogez, Steve A. Chien, Kiri Wagstaff and David R. Thompson.

Proceedings of the 12th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS), 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuchs_autonomous_2014,
    title = {Autonomous {Onboard} {Surface} {Feature} {Detection} for {Flyby} {Missions}},
    booktitle = {Proceedings of the 12th {International} {Symposium} on {Artificial} {Intelligence}, {Robotics} and {Automation} in {Space} (i-{SAIRAS})},
    author = {Fuchs, Thomas J. and Bue, Brian D. and Castillo-Rogez, Julie and Chien, Steve A. and Wagstaff, Kiri and Thompson, David R.},
    year = {2014},
}

Download Endnote/RIS citation

TY - CONF
TI - Autonomous Onboard Surface Feature Detection for Flyby Missions
AU - Fuchs, Thomas J.
AU - Bue, Brian D.
AU - Castillo-Rogez, Julie
AU - Chien, Steve A.
AU - Wagstaff, Kiri
AU - Thompson, David R.
C3 - Proceedings of the 12th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS)
DA - 2014///
PY - 2014
ER -

TMARKER: A Free Software Toolkit for Histopathological Cell Counting and Staining Estimation.

Peter J. Schüffler, Thomas J. Fuchs, Cheng S. Ong, Peter Wild and Joachim M. Buhmann.

Journal of Pathology Informatics, vol. 4, 2, p. 2, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{schuffler_tmarker_2013,
    title = {{TMARKER}: {A} {Free} {Software} {Toolkit} for {Histopathological} {Cell} {Counting} and {Staining} {Estimation}},
    volume = {4},
    url = {https://www.sciencedirect.com/science/article/pii/S2153353922006629?via%3Dihub},
    doi = {10.4103/2153-3539.109804},
    abstract = {Background: Histological tissue analysis often involves manual cell counting and staining estimation of cancerous cells. These assessments are extremely time consuming, highly subjective and prone to error, since immunohistochemically stained cancer tissues usually show high variability in cell sizes, morphological structures and staining quality. To facilitate reproducible analysis in clinical practice as well as for cancer research, objective computer assisted staining estimation is highly desirable. Methods: We employ machine learning algorithms as randomized decision trees and support vector machines for nucleus detection and classification. Superpixels as segmentation over the tissue image are classified into foreground and background and thereafter into malignant and benign, learning from the user's feedback. As a fast alternative without nucleus classification, the existing color deconvolution method is incorporated. Results: Our program TMARKER connects already available workflows for computational pathology and immunohistochemical tissue rating with modern active learning algorithms from machine learning and computer vision. On a test dataset of human renal clear cell carcinoma and prostate carcinoma, the performance of the used algorithms is equivalent to two independent pathologists for nucleus detection and classification. Conclusion: We present a novel, free and operating system independent software package for computational cell counting and staining estimation, supporting IHC stained tissue analysis in clinic and for research. Proprietary toolboxes for similar tasks are expensive, bound to specific commercial hardware (e.g. a microscope) and mostly not quantitatively validated in terms of performance and reproducibility. We are confident that the presented software package will proof valuable for the scientific community and we anticipate a broader application domain due to the possibility to interactively learn models for new image types.},
    number = {2},
    journal = {Journal of Pathology Informatics},
    author = {Sch\"uffler, Peter J. and Fuchs, Thomas J. and Ong, Cheng S. and Wild, Peter and Buhmann, Joachim M.},
    year = {2013},
    pages = {2},
}

Download Endnote/RIS citation

TY - JOUR
TI - TMARKER: A Free Software Toolkit for Histopathological Cell Counting and Staining Estimation
AU - Schüffler, Peter J.
AU - Fuchs, Thomas J.
AU - Ong, Cheng S.
AU - Wild, Peter
AU - Buhmann, Joachim M.
T2 - Journal of Pathology Informatics
AB - Background: Histological tissue analysis often involves manual cell counting and staining estimation of cancerous cells. These assessments are extremely time consuming, highly subjective and prone to error, since immunohistochemically stained cancer tissues usually show high variability in cell sizes, morphological structures and staining quality. To facilitate reproducible analysis in clinical practice as well as for cancer research, objective computer assisted staining estimation is highly desirable. Methods: We employ machine learning algorithms as randomized decision trees and support vector machines for nucleus detection and classification. Superpixels as segmentation over the tissue image are classified into foreground and background and thereafter into malignant and benign, learning from the user's feedback. As a fast alternative without nucleus classification, the existing color deconvolution method is incorporated. Results: Our program TMARKER connects already available workflows for computational pathology and immunohistochemical tissue rating with modern active learning algorithms from machine learning and computer vision. On a test dataset of human renal clear cell carcinoma and prostate carcinoma, the performance of the used algorithms is equivalent to two independent pathologists for nucleus detection and classification. Conclusion: We present a novel, free and operating system independent software package for computational cell counting and staining estimation, supporting IHC stained tissue analysis in clinic and for research. Proprietary toolboxes for similar tasks are expensive, bound to specific commercial hardware (e.g. a microscope) and mostly not quantitatively validated in terms of performance and reproducibility. We are confident that the presented software package will proof valuable for the scientific community and we anticipate a broader application domain due to the possibility to interactively learn models for new image types.
DA - 2013///
PY - 2013
DO - 10.4103/2153-3539.109804
VL - 4
IS - 2
SP - 2
UR - https://www.sciencedirect.com/science/article/pii/S2153353922006629?via%3Dihub
ER -

Background: Histological tissue analysis often involves manual cell counting and staining estimation of cancerous cells. These assessments are extremely time consuming, highly subjective and prone to error, since immunohistochemically stained cancer tissues usually show high variability in cell sizes, morphological structures and staining quality. To facilitate reproducible analysis in clinical practice as well as for cancer research, objective computer assisted staining estimation is highly desirable. Methods: We employ machine learning algorithms as randomized decision trees and support vector machines for nucleus detection and classification. Superpixels as segmentation over the tissue image are classified into foreground and background and thereafter into malignant and benign, learning from the user's feedback. As a fast alternative without nucleus classification, the existing color deconvolution method is incorporated. Results: Our program TMARKER connects already available workflows for computational pathology and immunohistochemical tissue rating with modern active learning algorithms from machine learning and computer vision. On a test dataset of human renal clear cell carcinoma and prostate carcinoma, the performance of the used algorithms is equivalent to two independent pathologists for nucleus detection and classification. Conclusion: We present a novel, free and operating system independent software package for computational cell counting and staining estimation, supporting IHC stained tissue analysis in clinic and for research. Proprietary toolboxes for similar tasks are expensive, bound to specific commercial hardware (e.g. a microscope) and mostly not quantitatively validated in terms of performance and reproducibility. We are confident that the presented software package will proof valuable for the scientific community and we anticipate a broader application domain due to the possibility to interactively learn models for new image types.

Smart, Texture-Sensitive Instrument Classification for in Situ Rock and Layer Analysis.

K. L. Wagstaff, D. R. Thompson, W. Abbey, A. Allwood, D. Bekker, N. A. Cabrol, Thomas J. Fuchs and K. Ortega.

Geophysical Research Letters, vol. 40, 16, p. 4188–4193, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{wagstaff_smart_2013,
    title = {Smart, {Texture}-{Sensitive} {Instrument} {Classification} for in {Situ} {Rock} and {Layer} {Analysis}},
    volume = {40},
    issn = {1944-8007},
    url = {http://dx.doi.org/10.1002/grl.50817},
    doi = {10.1002/grl.50817},
    number = {16},
    journal = {Geophysical Research Letters},
    author = {Wagstaff, K. L. and Thompson, D. R. and Abbey, W. and Allwood, A. and Bekker, D. and Cabrol, N. A. and Fuchs, Thomas J. and Ortega, K.},
    year = {2013},
    pages = {4188--4193},
}

Download Endnote/RIS citation

TY - JOUR
TI - Smart, Texture-Sensitive Instrument Classification for in Situ Rock and Layer Analysis
AU - Wagstaff, K. L.
AU - Thompson, D. R.
AU - Abbey, W.
AU - Allwood, A.
AU - Bekker, D.
AU - Cabrol, N. A.
AU - Fuchs, Thomas J.
AU - Ortega, K.
T2 - Geophysical Research Letters
DA - 2013///
PY - 2013
DO - 10.1002/grl.50817
VL - 40
IS - 16
SP - 4188
EP - 4193
SN - 1944-8007
UR - http://dx.doi.org/10.1002/grl.50817
ER -

TMARKER: A Robust and Free Software Toolkit for Histopathological Cell Counting and Immunohistochemical Staining Estimations.

Peter J. Schueffler, Niels Rupp, Cheng S. Ong, Joachim M. Buhmann, Thomas J. Fuchs and Peter J. Wild.

German Society of Pathology 97th Annual Meeting, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schueffler_tmarker:_2013,
    title = {{TMARKER}: {A} {Robust} and {Free} {Software} {Toolkit} for {Histopathological} {Cell} {Counting} and {Immunohistochemical} {Staining} {Estimations}},
    url = {http://link.springer.com/article/10.1007%2Fs00292-013-1765-2},
    abstract = {Aims. Assessment of immunhistochemical staining intensity and percentage of positive cells generally suffers from high inter- and intra-observer variability. Besides, estimation of the intensity and percentage of immunostained cells often involves manual cell counting and is therefore time consuming.
    Methods. A novel, free, and open-source software toolkit was developed, connecting already available work flows for computational pathology and immunohistochemical tissue assessment with modern active learning algorithms from machine learning and computer vision.
    Results. A new and platform independent software program, called TMARKER, was developed and introduced. The validity and robustness of the used algorithms has been shown on a test dataset of human renal clear cell carcinomas and prostate carcinomas.
    Conclusions. This new software program together with a user-friendly Java-based graphical user interface enabled comprehensive computational assistance in pathological tissue rating. Routine use of this software toolkit may provide more reliable and robust biomarkers for treatment decision.},
    booktitle = {German {Society} of {Pathology} 97th {Annual} {Meeting}},
    author = {Schueffler, Peter J. and Rupp, Niels and Ong, Cheng S. and Buhmann, Joachim M. and Fuchs, Thomas J. and Wild, Peter J.},
    year = {2013},
}

Download Endnote/RIS citation

TY - CONF
TI - TMARKER: A Robust and Free Software Toolkit for Histopathological Cell Counting and Immunohistochemical Staining Estimations
AU - Schueffler, Peter J.
AU - Rupp, Niels
AU - Ong, Cheng S.
AU - Buhmann, Joachim M.
AU - Fuchs, Thomas J.
AU - Wild, Peter J.
AB - Aims. Assessment of immunhistochemical staining intensity and percentage of positive cells generally suffers from high inter- and intra-observer variability. Besides, estimation of the intensity and percentage of immunostained cells often involves manual cell counting and is therefore time consuming.
Methods. A novel, free, and open-source software toolkit was developed, connecting already available work flows for computational pathology and immunohistochemical tissue assessment with modern active learning algorithms from machine learning and computer vision.
Results. A new and platform independent software program, called TMARKER, was developed and introduced. The validity and robustness of the used algorithms has been shown on a test dataset of human renal clear cell carcinomas and prostate carcinomas.
Conclusions. This new software program together with a user-friendly Java-based graphical user interface enabled comprehensive computational assistance in pathological tissue rating. Routine use of this software toolkit may provide more reliable and robust biomarkers for treatment decision.
C3 - German Society of Pathology 97th Annual Meeting
DA - 2013///
PY - 2013
UR - http://link.springer.com/article/10.1007%2Fs00292-013-1765-2
ER -

Aims. Assessment of immunhistochemical staining intensity and percentage of positive cells generally suffers from high inter- and intra-observer variability. Besides, estimation of the intensity and percentage of immunostained cells often involves manual cell counting and is therefore time consuming. Methods. A novel, free, and open-source software toolkit was developed, connecting already available work flows for computational pathology and immunohistochemical tissue assessment with modern active learning algorithms from machine learning and computer vision. Results. A new and platform independent software program, called TMARKER, was developed and introduced. The validity and robustness of the used algorithms has been shown on a test dataset of human renal clear cell carcinomas and prostate carcinomas. Conclusions. This new software program together with a user-friendly Java-based graphical user interface enabled comprehensive computational assistance in pathological tissue rating. Routine use of this software toolkit may provide more reliable and robust biomarkers for treatment decision.

Automated Analysis of Tissue Micro-Array Images on the Example of Renal Cell Carcinoma.

Peter J. Schueffler, Thomas J. Fuchs, Cheng Soon Ong, Volker Roth and Joachim M. Buhmann.

In: Similarity-Based Pattern Analysis and Recognition, p. 219–246, Advances in Computer Vision and Pattern Recognition, Springer, ISBN 978-1-4471-5627-7, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{schueffler_automated_2013,
    address = {London},
    series = {Advances in {Computer} {Vision} and {Pattern} {Recognition}},
    title = {Automated {Analysis} of {Tissue} {Micro}-{Array} {Images} on the {Example} of {Renal} {Cell} {Carcinoma}},
    isbn = {978-1-4471-5627-7},
    url = {https://www.springer.com/computer/image+processing/book/978-1-4471-5627-7},
    abstract = {Automated tissue micro-array analysis forms a challenging problem in computational pathology. The detection of cell nuclei, the classification into malignant and benign as well as the evaluation of their protein expression pattern by immunohistochemical staining are crucial routine steps for human cancer research and oncology. Computational assistance in this field can extremely accelerate the high throughput of the upcoming patient data as well as facilitate the reproducibility and objectivity of qualitative and quantitative measures. In this chapter, we describe an automated pipeline for staining estimation of tissue micro-array images, which comprises nucleus detection, nucleus segmentation, nucleus classification and staining estimation among cancerous nuclei. This pipeline is a practical example for the importance of non-metric effects in this kind of image analysis, e.g., the use of shape information and non-Euclidean kernels improve the nucleus classification performance significantly. The pipeline is explained and validated on a renal clear cell carcinoma dataset with MIB-1 stained tissue micro-array images and survival data of 133 patients. Further, the pipeline is implemented for medical use and research purpose in the free program TMARKER.},
    booktitle = {Similarity-{Based} {Pattern} {Analysis} and {Recognition}},
    publisher = {Springer},
    author = {Schueffler, Peter J. and Fuchs, Thomas J. and Ong, Cheng Soon and Roth, Volker and Buhmann, Joachim M.},
    year = {2013},
    pages = {219--246},
}

Download Endnote/RIS citation

TY - CHAP
TI - Automated Analysis of Tissue Micro-Array Images on the Example of Renal Cell Carcinoma
AU - Schueffler, Peter J.
AU - Fuchs, Thomas J.
AU - Ong, Cheng Soon
AU - Roth, Volker
AU - Buhmann, Joachim M.
T2 - Similarity-Based Pattern Analysis and Recognition
T3 - Advances in Computer Vision and Pattern Recognition
AB - Automated tissue micro-array analysis forms a challenging problem in computational pathology. The detection of cell nuclei, the classification into malignant and benign as well as the evaluation of their protein expression pattern by immunohistochemical staining are crucial routine steps for human cancer research and oncology. Computational assistance in this field can extremely accelerate the high throughput of the upcoming patient data as well as facilitate the reproducibility and objectivity of qualitative and quantitative measures. In this chapter, we describe an automated pipeline for staining estimation of tissue micro-array images, which comprises nucleus detection, nucleus segmentation, nucleus classification and staining estimation among cancerous nuclei. This pipeline is a practical example for the importance of non-metric effects in this kind of image analysis, e.g., the use of shape information and non-Euclidean kernels improve the nucleus classification performance significantly. The pipeline is explained and validated on a renal clear cell carcinoma dataset with MIB-1 stained tissue micro-array images and survival data of 133 patients. Further, the pipeline is implemented for medical use and research purpose in the free program TMARKER.
CY - London
DA - 2013///
PY - 2013
SP - 219
EP - 246
PB - Springer
SN - 978-1-4471-5627-7
UR - https://www.springer.com/computer/image+processing/book/978-1-4471-5627-7
ER -

Automated tissue micro-array analysis forms a challenging problem in computational pathology. The detection of cell nuclei, the classification into malignant and benign as well as the evaluation of their protein expression pattern by immunohistochemical staining are crucial routine steps for human cancer research and oncology. Computational assistance in this field can extremely accelerate the high throughput of the upcoming patient data as well as facilitate the reproducibility and objectivity of qualitative and quantitative measures. In this chapter, we describe an automated pipeline for staining estimation of tissue micro-array images, which comprises nucleus detection, nucleus segmentation, nucleus classification and staining estimation among cancerous nuclei. This pipeline is a practical example for the importance of non-metric effects in this kind of image analysis, e.g., the use of shape information and non-Euclidean kernels improve the nucleus classification performance significantly. The pipeline is explained and validated on a renal clear cell carcinoma dataset with MIB-1 stained tissue micro-array images and survival data of 133 patients. Further, the pipeline is implemented for medical use and research purpose in the free program TMARKER.

A Model Development Pipeline for Crohn's Disease Severity Assessment from Magnetic Resonance Images.

Peter J. Schüffler, Dwarikanath Mahapatra, Jeroen A. W. Tielbeek, Franciscus M. Vos, Jesica Makanyanga, Doug A. Pendsé, C. Yung Nio, Jaap Stoker, Stuart A. Taylor and Joachim M. Buhmann.

In: Hiroyuki Yoshida, Simon Warfield and Michael Vannier (eds.) Abdominal Imaging. Computation and Clinical Applications, vol. 8198, p. 1-10, Lecture Notes in Computer Science, Springer Berlin Heidelberg, ISBN 978-3-642-41082-6, 2013

Outstanding Paper Award

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{schuffler_model_2013,
    series = {Lecture {Notes} in {Computer} {Science}},
    title = {A {Model} {Development} {Pipeline} for {Crohn}'s {Disease} {Severity} {Assessment} from {Magnetic} {Resonance} {Images}},
    volume = {8198},
    isbn = {978-3-642-41082-6},
    url = {http://link.springer.com/chapter/10.1007%2F978-3-642-41083-3_1},
    abstract = {Crohn's Disease affects the intestinal tract of a patient and can have varying severity which influences treatment strategy. The clinical severity score CDEIS (Crohn’s Disease Endoscopic Index of severity) ranges from 0 to 44 and is measured by endoscopy. In this paper we investigate the potential of non-invasive magnetic resonance imaging to assess this severity, together with the underlying question which features are most relevant for this estimation task. We propose a new general and modular pipeline that uses machine learning techniques to quantify disease severity from MR images and show its value on Crohn’s Disease severity assessment on 30 patients scored by 4 medical experts. With the pipeline, we can obtain a magnetic resonance imaging score which outperforms two existing reference scores MaRIA and AIS.},
    booktitle = {Abdominal {Imaging}. {Computation} and {Clinical} {Applications}},
    publisher = {Springer Berlin Heidelberg},
    author = {Sch\"uffler, Peter J. and Mahapatra, Dwarikanath and Tielbeek, Jeroen A. W. and Vos, Franciscus M. and Makanyanga, Jesica and Pends\'e, Doug A. and Nio, C. Yung and Stoker, Jaap and Taylor, Stuart A. and Buhmann, Joachim M.},
    editor = {Yoshida, Hiroyuki and Warfield, Simon and Vannier, Michael},
    year = {2013},
    keywords = {AIS, CDEIS, Crohn’s Disease, MaRIA, abdominal MRI},
    pages = {1--10},
}

Download Endnote/RIS citation

TY - CHAP
TI - A Model Development Pipeline for Crohn's Disease Severity Assessment from Magnetic Resonance Images
AU - Schüffler, Peter J.
AU - Mahapatra, Dwarikanath
AU - Tielbeek, Jeroen A. W.
AU - Vos, Franciscus M.
AU - Makanyanga, Jesica
AU - Pendsé, Doug A.
AU - Nio, C. Yung
AU - Stoker, Jaap
AU - Taylor, Stuart A.
AU - Buhmann, Joachim M.
T2 - Abdominal Imaging. Computation and Clinical Applications
A2 - Yoshida, Hiroyuki
A2 - Warfield, Simon
A2 - Vannier, Michael
T3 - Lecture Notes in Computer Science
AB - Crohn's Disease affects the intestinal tract of a patient and can have varying severity which influences treatment strategy. The clinical severity score CDEIS (Crohn’s Disease Endoscopic Index of severity) ranges from 0 to 44 and is measured by endoscopy. In this paper we investigate the potential of non-invasive magnetic resonance imaging to assess this severity, together with the underlying question which features are most relevant for this estimation task. We propose a new general and modular pipeline that uses machine learning techniques to quantify disease severity from MR images and show its value on Crohn’s Disease severity assessment on 30 patients scored by 4 medical experts. With the pipeline, we can obtain a magnetic resonance imaging score which outperforms two existing reference scores MaRIA and AIS.
DA - 2013///
PY - 2013
VL - 8198
SP - 1
EP - 10
PB - Springer Berlin Heidelberg
SN - 978-3-642-41082-6
UR - http://link.springer.com/chapter/10.1007%2F978-3-642-41083-3_1
KW - AIS
KW - CDEIS
KW - Crohn’s Disease
KW - MaRIA
KW - abdominal MRI
ER -

Crohn's Disease affects the intestinal tract of a patient and can have varying severity which influences treatment strategy. The clinical severity score CDEIS (Crohn’s Disease Endoscopic Index of severity) ranges from 0 to 44 and is measured by endoscopy. In this paper we investigate the potential of non-invasive magnetic resonance imaging to assess this severity, together with the underlying question which features are most relevant for this estimation task. We propose a new general and modular pipeline that uses machine learning techniques to quantify disease severity from MR images and show its value on Crohn’s Disease severity assessment on 30 patients scored by 4 medical experts. With the pipeline, we can obtain a magnetic resonance imaging score which outperforms two existing reference scores MaRIA and AIS.

Feature Selection Strategies for Classifying High Dimensional Astronomical Data Sets.

Ciro Donalek, Arun Kumar A., S. G. Djorgovski, Ashish A. Mahabal, Matthew J. Graham, Thomas J. Fuchs, Michael J. Turmon, N. Sajeeth Philip, Michael Ting-Chang Yang and Giuseppe Longo.

IEEE International Conference on Big Data, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{donalek_feature_2013,
    title = {Feature {Selection} {Strategies} for {Classifying} {High} {Dimensional} {Astronomical} {Data} {Sets}},
    url = {http://arxiv.org/abs/1310.1976},
    abstract = {The amount of collected data in many scientific fields is increasing, all of them requiring a common task: extract knowledge from massive, multi parametric data sets, as rapidly and efficiently possible. This is especially true in astronomy where synoptic sky surveys are enabling new research frontiers in the time domain astronomy and posing several new object classification challenges in multi dimensional spaces; given the high number of parameters available for each object, feature selection is quickly becoming a crucial task in analyzing astronomical data sets. Using data sets extracted from the ongoing Catalina Real-Time Transient Surveys (CRTS) and the Kepler Mission we illustrate a variety of feature selection strategies used to identify the subsets that give the most information and the results achieved applying these techniques to three major astronomical problems.},
    urldate = {2016-01-05},
    journal = {IEEE International Conference on Big Data},
    author = {Donalek, Ciro and A., Arun Kumar and Djorgovski, S. G. and Mahabal, Ashish A. and Graham, Matthew J. and Fuchs, Thomas J. and Turmon, Michael J. and Philip, N. Sajeeth and Yang, Michael Ting-Chang and Longo, Giuseppe},
    year = {2013},
    keywords = {Astrophysics - Instrumentation and Methods for Astrophysics, Computer Science - Computer Vision and Pattern Recognition},
}

Download Endnote/RIS citation

TY - JOUR
TI - Feature Selection Strategies for Classifying High Dimensional Astronomical Data Sets
AU - Donalek, Ciro
AU - A., Arun Kumar
AU - Djorgovski, S. G.
AU - Mahabal, Ashish A.
AU - Graham, Matthew J.
AU - Fuchs, Thomas J.
AU - Turmon, Michael J.
AU - Philip, N. Sajeeth
AU - Yang, Michael Ting-Chang
AU - Longo, Giuseppe
T2 - IEEE International Conference on Big Data
AB - The amount of collected data in many scientific fields is increasing, all of them requiring a common task: extract knowledge from massive, multi parametric data sets, as rapidly and efficiently possible. This is especially true in astronomy where synoptic sky surveys are enabling new research frontiers in the time domain astronomy and posing several new object classification challenges in multi dimensional spaces; given the high number of parameters available for each object, feature selection is quickly becoming a crucial task in analyzing astronomical data sets. Using data sets extracted from the ongoing Catalina Real-Time Transient Surveys (CRTS) and the Kepler Mission we illustrate a variety of feature selection strategies used to identify the subsets that give the most information and the results achieved applying these techniques to three major astronomical problems.
DA - 2013///
PY - 2013
DP - arXiv.org
UR - http://arxiv.org/abs/1310.1976
Y2 - 2016/01/05/04:03:47
KW - Astrophysics - Instrumentation and Methods for Astrophysics
KW - Computer Science - Computer Vision and Pattern Recognition
ER -

The amount of collected data in many scientific fields is increasing, all of them requiring a common task: extract knowledge from massive, multi parametric data sets, as rapidly and efficiently possible. This is especially true in astronomy where synoptic sky surveys are enabling new research frontiers in the time domain astronomy and posing several new object classification challenges in multi dimensional spaces; given the high number of parameters available for each object, feature selection is quickly becoming a crucial task in analyzing astronomical data sets. Using data sets extracted from the ongoing Catalina Real-Time Transient Surveys (CRTS) and the Kepler Mission we illustrate a variety of feature selection strategies used to identify the subsets that give the most information and the results achieved applying these techniques to three major astronomical problems.

Quickly Boosting Decision Trees – Pruning Underachieving Features using a Provable Bound.

Ron Appel, Piotr Dollar, Thomas J. Fuchs and Pietro Perona.

Proceedings of the 30th International Conference on Machine Learning (ICML), vol. 28, p. 594-602, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{appel_quickly_2013,
    title = {Quickly {Boosting} {Decision} {Trees} – {Pruning} {Underachieving} {Features} using a {Provable} {Bound}},
    volume = {28},
    url = {http://jmlr.org/proceedings/papers/v28/appel13.html},
    booktitle = {Proceedings of the 30th {International} {Conference} on {Machine} {Learning} ({ICML})},
    author = {Appel, Ron and Dollar, Piotr and Fuchs, Thomas J. and Perona, Pietro},
    year = {2013},
    pages = {594--602},
}

Download Endnote/RIS citation

TY - CONF
TI - Quickly Boosting Decision Trees – Pruning Underachieving Features using a Provable Bound
AU - Appel, Ron
AU - Dollar, Piotr
AU - Fuchs, Thomas J.
AU - Perona, Pietro
C3 - Proceedings of the 30th International Conference on Machine Learning (ICML)
DA - 2013///
PY - 2013
VL - 28
SP - 594
EP - 602
UR - http://jmlr.org/proceedings/papers/v28/appel13.html
ER -

Recognizing Humans in Motion: Trajectory-based Areal Video Analysis.

Yumi Iwashita, Michael Ryoo, Thomas J. Fuchs and Curtis Padgett.

24th British Machine Vision Conference (BMVC), 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{iwashita_recognizing_2013,
    title = {Recognizing {Humans} in {Motion}: {Trajectory}-based {Areal} {Video} {Analysis}},
    booktitle = {24th {British} {Machine} {Vision} {Conference} ({BMVC})},
    author = {Iwashita, Yumi and Ryoo, Michael and Fuchs, Thomas J. and Padgett, Curtis},
    year = {2013},
}

Download Endnote/RIS citation

TY - CONF
TI - Recognizing Humans in Motion: Trajectory-based Areal Video Analysis
AU - Iwashita, Yumi
AU - Ryoo, Michael
AU - Fuchs, Thomas J.
AU - Padgett, Curtis
C3 - 24th British Machine Vision Conference (BMVC)
DA - 2013///
PY - 2013
ER -

Automatic Detection and Segmentation of Crohn's Disease Tissues from Abdominal MRI.

D. Mahapatra, P. J. Schüffler, J. A. W. Tielbeek, J. C. Makanyanga, J. Stoker, S. A. Taylor, F. M. Vos and J. M. Buhmann.

IEEE Transactions on Medical Imaging, vol. 32, p. 2332-2347, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{mahapatra_automatic_2013,
    title = {Automatic {Detection} and {Segmentation} of {Crohn}'s {Disease} {Tissues} from {Abdominal} {MRI}},
    volume = {32},
    issn = {0278-0062},
    url = {http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6600949},
    doi = {10.1109/TMI.2013.2282124},
    abstract = {We propose an information processing pipeline for segmenting parts of the bowel in abdominal magnetic resonance images that are affected with Crohn's disease. Given a magnetic resonance imaging test volume, it is first oversegmented into supervoxels and each supervoxel is analyzed to detect presence of Crohn's disease using random forest (RF) classifiers. The supervoxels identified as containing diseased tissues define the volume of interest (VOI). All voxels within the VOI are further investigated to segment the diseased region. Probability maps are generated for each voxel using a second set of RF classifiers which give the probabilities of each voxel being diseased, normal or background. The negative log-likelihood of these maps are used as penalty costs in a graph cut segmentation framework. Low level features like intensity statistics, texture anisotropy and curvature asymmetry, and high level context features are used at different stages. Smoothness constraints are imposed based on semantic information (importance of each feature to the classification task) derived from the second set of learned RF classifiers. Experimental results show that our method achieves high segmentation accuracy with Dice metric values of 0.90 ± 0.04 and Hausdorff distance of 7.3 ± 0.8 mm. Semantic information and context features are an integral part of our method and are robust to different levels of added noise.},
    journal = {IEEE Transactions on Medical Imaging},
    author = {Mahapatra, D. and Sch\"uffler, P. J. and Tielbeek, J. A. W. and Makanyanga, J. C. and Stoker, J. and Taylor, S. A. and Vos, F. M. and Buhmann, J. M.},
    year = {2013},
    keywords = {Anisotropic magnetoresistance, Context, Crohn\&\#x2019, Diseases, Entropy, Image segmentation, Radio frequency, content, graph cut, image features, probability maps, random forests, s disease, segmentation, semantic information, shape, supervoxels},
    pages = {2332--2347},
}

Download Endnote/RIS citation

TY - JOUR
TI - Automatic Detection and Segmentation of Crohn's Disease Tissues from Abdominal MRI
AU - Mahapatra, D.
AU - Schüffler, P. J.
AU - Tielbeek, J. A. W.
AU - Makanyanga, J. C.
AU - Stoker, J.
AU - Taylor, S. A.
AU - Vos, F. M.
AU - Buhmann, J. M.
T2 - IEEE Transactions on Medical Imaging
AB - We propose an information processing pipeline for segmenting parts of the bowel in abdominal magnetic resonance images that are affected with Crohn's disease. Given a magnetic resonance imaging test volume, it is first oversegmented into supervoxels and each supervoxel is analyzed to detect presence of Crohn's disease using random forest (RF) classifiers. The supervoxels identified as containing diseased tissues define the volume of interest (VOI). All voxels within the VOI are further investigated to segment the diseased region. Probability maps are generated for each voxel using a second set of RF classifiers which give the probabilities of each voxel being diseased, normal or background. The negative log-likelihood of these maps are used as penalty costs in a graph cut segmentation framework. Low level features like intensity statistics, texture anisotropy and curvature asymmetry, and high level context features are used at different stages. Smoothness constraints are imposed based on semantic information (importance of each feature to the classification task) derived from the second set of learned RF classifiers. Experimental results show that our method achieves high segmentation accuracy with Dice metric values of 0.90 ± 0.04 and Hausdorff distance of 7.3 ± 0.8 mm. Semantic information and context features are an integral part of our method and are robust to different levels of added noise.
DA - 2013///
PY - 2013
DO - 10.1109/TMI.2013.2282124
VL - 32
SP - 2332
EP - 2347
SN - 0278-0062
UR - http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6600949
KW - Anisotropic magnetoresistance
KW - Context
KW - Crohn’
KW - Diseases
KW - Entropy
KW - Image segmentation
KW - Radio frequency
KW - content
KW - graph cut
KW - image features
KW - probability maps
KW - random forests
KW - s disease
KW - segmentation
KW - semantic information
KW - shape
KW - supervoxels
ER -

We propose an information processing pipeline for segmenting parts of the bowel in abdominal magnetic resonance images that are affected with Crohn's disease. Given a magnetic resonance imaging test volume, it is first oversegmented into supervoxels and each supervoxel is analyzed to detect presence of Crohn's disease using random forest (RF) classifiers. The supervoxels identified as containing diseased tissues define the volume of interest (VOI). All voxels within the VOI are further investigated to segment the diseased region. Probability maps are generated for each voxel using a second set of RF classifiers which give the probabilities of each voxel being diseased, normal or background. The negative log-likelihood of these maps are used as penalty costs in a graph cut segmentation framework. Low level features like intensity statistics, texture anisotropy and curvature asymmetry, and high level context features are used at different stages. Smoothness constraints are imposed based on semantic information (importance of each feature to the classification task) derived from the second set of learned RF classifiers. Experimental results show that our method achieves high segmentation accuracy with Dice metric values of 0.90 ± 0.04 and Hausdorff distance of 7.3 ± 0.8 mm. Semantic information and context features are an integral part of our method and are robust to different levels of added noise.

TextureCam: A Smart Camera for Microscale, Mesoscale, and Deep Space.

William Abbey, Abigail Allwood, Dmitriy Bekker, Benjamin Bornstein, Nathalie A. Cabrol, Rebecca Castano, Steve A. Chien, Joshua Doubleday, Tara Estlin, Greydon Foil, Thomas J. Fuchs, Daniel Howarth, Kevin Ortega, David R. Thompson and Kiri L. Wagstaff.

44th Lunar and Planetary Science Conference, p. 2209, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{abbey_texturecam:_2013,
    title = {{TextureCam}: {A} {Smart} {Camera} for {Microscale}, {Mesoscale}, and {Deep} {Space}},
    booktitle = {44th {Lunar} and {Planetary} {Science} {Conference}},
    author = {Abbey, William and Allwood, Abigail and Bekker, Dmitriy and Bornstein, Benjamin and Cabrol, Nathalie A. and Castano, Rebecca and Chien, Steve A. and Doubleday, Joshua and Estlin, Tara and Foil, Greydon and Fuchs, Thomas J. and Howarth, Daniel and Ortega, Kevin and Thompson, David R. and Wagstaff, Kiri L.},
    year = {2013},
    pages = {2209},
}

Download Endnote/RIS citation

TY - CONF
TI - TextureCam: A Smart Camera for Microscale, Mesoscale, and Deep Space
AU - Abbey, William
AU - Allwood, Abigail
AU - Bekker, Dmitriy
AU - Bornstein, Benjamin
AU - Cabrol, Nathalie A.
AU - Castano, Rebecca
AU - Chien, Steve A.
AU - Doubleday, Joshua
AU - Estlin, Tara
AU - Foil, Greydon
AU - Fuchs, Thomas J.
AU - Howarth, Daniel
AU - Ortega, Kevin
AU - Thompson, David R.
AU - Wagstaff, Kiri L.
C3 - 44th Lunar and Planetary Science Conference
DA - 2013///
PY - 2013
SP - 2209
ER -

Structure Preserving Embedding of Dissimilarity Data.

Volker Roth, Thomas J. Fuchs, Julia E. Vogt, Sandhya Prabhakaran and Joachim M. Buhmann.

In: Similarity-Based Pattern Analysis and Recognition, p. 157–178, Advances in Computer Vision and Pattern Recognition, Springer, ISBN 978-1-4471-5627-7, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@incollection{roth_structure_2013,
    series = {Advances in {Computer} {Vision} and {Pattern} {Recognition}},
    title = {Structure {Preserving} {Embedding} of {Dissimilarity} {Data}},
    isbn = {978-1-4471-5627-7},
    url = {https://www.springer.com/computer/image+processing/book/978-1-4471-5627-7},
    booktitle = {Similarity-{Based} {Pattern} {Analysis} and {Recognition}},
    publisher = {Springer},
    author = {Roth, Volker and Fuchs, Thomas J. and Vogt, Julia E. and Prabhakaran, Sandhya and Buhmann, Joachim M.},
    year = {2013},
    pages = {157--178},
}

Download Endnote/RIS citation

TY - CHAP
TI - Structure Preserving Embedding of Dissimilarity Data
AU - Roth, Volker
AU - Fuchs, Thomas J.
AU - Vogt, Julia E.
AU - Prabhakaran, Sandhya
AU - Buhmann, Joachim M.
T2 - Similarity-Based Pattern Analysis and Recognition
T3 - Advances in Computer Vision and Pattern Recognition
DA - 2013///
PY - 2013
SP - 157
EP - 178
PB - Springer
SN - 978-1-4471-5627-7
UR - https://www.springer.com/computer/image+processing/book/978-1-4471-5627-7
ER -

p53 Suppresses Type II Endometrial Carcinomas in Mice and Governs Endometrial Tumour Aggressiveness in Humans.

Peter J. Wild, Kristian Ikenberg, Thomas J. Fuchs, Markus Rechsteiner, Strahil Georgiev, Niklaus Fankhauser, Aurelia Noske, Matthias Roessle, Rosmarie Caduff, Athanassios Dellas, Daniel Fink, Holger Moch, Wilhelm Krek and Ian J. Frew.

EMBO Molecular Medicine, vol. 4, 8, p. 808-824, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{wild_p53_2012,
    title = {p53 {Suppresses} {Type} {II} {Endometrial} {Carcinomas} in {Mice} and {Governs} {Endometrial} {Tumour} {Aggressiveness} in {Humans}},
    volume = {4},
    issn = {17574676},
    shorttitle = {p53 suppresses type {II} endometrial carcinomas in mice and governs endometrial tumour aggressiveness in humans},
    url = {http://embomolmed.embopress.org/cgi/doi/10.1002/emmm.201101063},
    doi = {10.1002/emmm.201101063},
    abstract = {Type II endometrial carcinomas are a highly aggressive group of tumour subtypes that are frequently associated with inactivation of the TP53 tumour suppressor gene. We show that mice with endometrium-specific deletion of Trp53 initially exhibited histological changes that are identical to known precursor lesions of type II endometrial carcinomas in humans and later developed carcinomas representing all type II subtypes. The mTORC1 signalling pathway was frequently activated in these precursor lesions and tumours, suggesting a genetic cooperation between this pathway and Trp53 deficiency in tumour initiation. Consistent with this idea, analyses of 521 human endometrial carcinomas identified frequent mTORC1 pathway activation in type I as well as type II endometrial carcinoma subtypes. mTORC1 pathway activation and p53 expression or mutation status each independently predicted poor patient survival. We suggest that molecular alterations in p53 and the mTORC1 pathway play different roles in the initiation of the different endometrial cancer subtypes, but that combined p53 inactivation and mTORC1 pathway activation are unifying pathogenic features among histologically diverse subtypes of late stage aggressive endometrial tumours.},
    language = {en},
    number = {8},
    urldate = {2016-04-12},
    journal = {EMBO Molecular Medicine},
    author = {Wild, Peter J. and Ikenberg, Kristian and Fuchs, Thomas J. and Rechsteiner, Markus and Georgiev, Strahil and Fankhauser, Niklaus and Noske, Aurelia and Roessle, Matthias and Caduff, Rosmarie and Dellas, Athanassios and Fink, Daniel and Moch, Holger and Krek, Wilhelm and Frew, Ian J.},
    year = {2012},
    pages = {808--824},
}

Download Endnote/RIS citation

TY - JOUR
TI - p53 Suppresses Type II Endometrial Carcinomas in Mice and Governs Endometrial Tumour Aggressiveness in Humans
AU - Wild, Peter J.
AU - Ikenberg, Kristian
AU - Fuchs, Thomas J.
AU - Rechsteiner, Markus
AU - Georgiev, Strahil
AU - Fankhauser, Niklaus
AU - Noske, Aurelia
AU - Roessle, Matthias
AU - Caduff, Rosmarie
AU - Dellas, Athanassios
AU - Fink, Daniel
AU - Moch, Holger
AU - Krek, Wilhelm
AU - Frew, Ian J.
T2 - EMBO Molecular Medicine
AB - Type II endometrial carcinomas are a highly aggressive group of tumour subtypes that are frequently associated with inactivation of the TP53 tumour suppressor gene. We show that mice with endometrium-specific deletion of Trp53 initially exhibited histological changes that are identical to known precursor lesions of type II endometrial carcinomas in humans and later developed carcinomas representing all type II subtypes. The mTORC1 signalling pathway was frequently activated in these precursor lesions and tumours, suggesting a genetic cooperation between this pathway and Trp53 deficiency in tumour initiation. Consistent with this idea, analyses of 521 human endometrial carcinomas identified frequent mTORC1 pathway activation in type I as well as type II endometrial carcinoma subtypes. mTORC1 pathway activation and p53 expression or mutation status each independently predicted poor patient survival. We suggest that molecular alterations in p53 and the mTORC1 pathway play different roles in the initiation of the different endometrial cancer subtypes, but that combined p53 inactivation and mTORC1 pathway activation are unifying pathogenic features among histologically diverse subtypes of late stage aggressive endometrial tumours.
DA - 2012///
PY - 2012
DO - 10.1002/emmm.201101063
DP - CrossRef
VL - 4
IS - 8
SP - 808
EP - 824
LA - en
SN - 17574676
ST - p53 suppresses type II endometrial carcinomas in mice and governs endometrial tumour aggressiveness in humans
UR - http://embomolmed.embopress.org/cgi/doi/10.1002/emmm.201101063
Y2 - 2016/04/12/13:58:18
ER -

Type II endometrial carcinomas are a highly aggressive group of tumour subtypes that are frequently associated with inactivation of the TP53 tumour suppressor gene. We show that mice with endometrium-specific deletion of Trp53 initially exhibited histological changes that are identical to known precursor lesions of type II endometrial carcinomas in humans and later developed carcinomas representing all type II subtypes. The mTORC1 signalling pathway was frequently activated in these precursor lesions and tumours, suggesting a genetic cooperation between this pathway and Trp53 deficiency in tumour initiation. Consistent with this idea, analyses of 521 human endometrial carcinomas identified frequent mTORC1 pathway activation in type I as well as type II endometrial carcinoma subtypes. mTORC1 pathway activation and p53 expression or mutation status each independently predicted poor patient survival. We suggest that molecular alterations in p53 and the mTORC1 pathway play different roles in the initiation of the different endometrial cancer subtypes, but that combined p53 inactivation and mTORC1 pathway activation are unifying pathogenic features among histologically diverse subtypes of late stage aggressive endometrial tumours.

End-to-End Dexterous Manipulation with Deliberate Interactive Estimation.

Nicolas H. Hudson, Tom Howard, Jeremy Ma, Abhinandan Jain, Max Bajracharya, Steven Myint, Larry Matthies, Paul Backes, Paul Hebert, Thomas J. Fuchs and Joel Burdick.

IEEE International Conference on Robotics and Automation (ICRA), p. 2371-2378, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{hudson_end--end_2012,
    title = {End-to-{End} {Dexterous} {Manipulation} with {Deliberate} {Interactive} {Estimation}},
    url = {http://ieeexplore.ieee.org/xpl/abstractKeywords.jsp?reload=true&arnumber=6225101&contentType=Conference+Publications},
    doi = {10.1109/ICRA.2012.6225101},
    abstract = {This paper presents a model based approach to autonomous dexterous manipulation, developed as part of the DARPA Autonomous Robotic Manipulation (ARM) program. The developed autonomy system uses robot, object, and environment models to identify and localize objects, and well as plan and execute required manipulation tasks. Deliberate interaction with objects and the environment increases system knowledge about the combined robot and environmental state, enabling high precision tasks such as key insertion to be performed in a consistent framework. This approach has been demonstrated across a wide range of manipulation tasks, and in independent DARPA testing archived the most successfully completed tasks with the fastest average task execution of any evaluated team.},
    booktitle = {{IEEE} {International} {Conference} on {Robotics} and {Automation} ({ICRA})},
    author = {Hudson, Nicolas H. and Howard, Tom and Ma, Jeremy and Jain, Abhinandan and Bajracharya, Max and Myint, Steven and Matthies, Larry and Backes, Paul and Hebert, Paul and Fuchs, Thomas J. and Burdick, Joel},
    year = {2012},
    pages = {2371--2378},
}

Download Endnote/RIS citation

TY - CONF
TI - End-to-End Dexterous Manipulation with Deliberate Interactive Estimation
AU - Hudson, Nicolas H.
AU - Howard, Tom
AU - Ma, Jeremy
AU - Jain, Abhinandan
AU - Bajracharya, Max
AU - Myint, Steven
AU - Matthies, Larry
AU - Backes, Paul
AU - Hebert, Paul
AU - Fuchs, Thomas J.
AU - Burdick, Joel
AB - This paper presents a model based approach to autonomous dexterous manipulation, developed as part of the DARPA Autonomous Robotic Manipulation (ARM) program. The developed autonomy system uses robot, object, and environment models to identify and localize objects, and well as plan and execute required manipulation tasks. Deliberate interaction with objects and the environment increases system knowledge about the combined robot and environmental state, enabling high precision tasks such as key insertion to be performed in a consistent framework. This approach has been demonstrated across a wide range of manipulation tasks, and in independent DARPA testing archived the most successfully completed tasks with the fastest average task execution of any evaluated team.
C3 - IEEE International Conference on Robotics and Automation (ICRA)
DA - 2012///
PY - 2012
DO - 10.1109/ICRA.2012.6225101
SP - 2371
EP - 2378
UR - http://ieeexplore.ieee.org/xpl/abstractKeywords.jsp?reload=true&arnumber=6225101&contentType=Conference+Publications
ER -

This paper presents a model based approach to autonomous dexterous manipulation, developed as part of the DARPA Autonomous Robotic Manipulation (ARM) program. The developed autonomy system uses robot, object, and environment models to identify and localize objects, and well as plan and execute required manipulation tasks. Deliberate interaction with objects and the environment increases system knowledge about the combined robot and environmental state, enabling high precision tasks such as key insertion to be performed in a consistent framework. This approach has been demonstrated across a wide range of manipulation tasks, and in independent DARPA testing archived the most successfully completed tasks with the fastest average task execution of any evaluated team.

Smart Cameras for Remote Science Survey.

David R. Thompson, William Abbey, Abigail Allwood, Dmitriy Bekker, Benjamin Bornstein, Nathalie A. Cabrol, Rebecca Castano, Tara Estlin, Thomas J. Fuchs and Kiri L. Wagstaff.

Proceedings of the 10th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS), 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{thompson_smart_2012,
    title = {Smart {Cameras} for {Remote} {Science} {Survey}},
    url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.308.5807},
    abstract = {Communication with remote exploration spacecraft is often intermittent and bandwidth is highly constrained. Future missions could use onboard science data understanding to prioritize downlink of critical features [1], draft summary maps of visited terrain [2], or identify targets of opportunity for followup measurements [3]. We describe a generic approach to classify geologic surfaces for autonomous science operations, suitable for parallelized implementations in FPGA hardware. We map these surfaces with texture channels- distinctive numerical signatures that differentiate properties such as roughness, pavement coatings, regolith characteristics, sedimentary fabrics and differential outcrop weathering. This work describes our basic image analysis approach and reports an initial performance evaluation using surface images from the Mars Exploration Rovers. Future work will incorporate these methods into camera hardware for real-time processing.},
    booktitle = {Proceedings of the 10th {International} {Symposium} on {Artificial} {Intelligence}, {Robotics} and {Automation} in {Space} (i-{SAIRAS})},
    author = {Thompson, David R. and Abbey, William and Allwood, Abigail and Bekker, Dmitriy and Bornstein, Benjamin and Cabrol, Nathalie A. and Castano, Rebecca and Estlin, Tara and Fuchs, Thomas J. and Wagstaff, Kiri L.},
    year = {2012},
}

Download Endnote/RIS citation

TY - CONF
TI - Smart Cameras for Remote Science Survey
AU - Thompson, David R.
AU - Abbey, William
AU - Allwood, Abigail
AU - Bekker, Dmitriy
AU - Bornstein, Benjamin
AU - Cabrol, Nathalie A.
AU - Castano, Rebecca
AU - Estlin, Tara
AU - Fuchs, Thomas J.
AU - Wagstaff, Kiri L.
AB - Communication with remote exploration spacecraft is often intermittent and bandwidth is highly constrained. Future missions could use onboard science data understanding to prioritize downlink of critical features [1], draft summary maps of visited terrain [2], or identify targets of opportunity for followup measurements [3]. We describe a generic approach to classify geologic surfaces for autonomous science operations, suitable for parallelized implementations in FPGA hardware. We map these surfaces with texture channels- distinctive numerical signatures that differentiate properties such as roughness, pavement coatings, regolith characteristics, sedimentary fabrics and differential outcrop weathering. This work describes our basic image analysis approach and reports an initial performance evaluation using surface images from the Mars Exploration Rovers. Future work will incorporate these methods into camera hardware for real-time processing.
C3 - Proceedings of the 10th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS)
DA - 2012///
PY - 2012
UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.308.5807
ER -

Communication with remote exploration spacecraft is often intermittent and bandwidth is highly constrained. Future missions could use onboard science data understanding to prioritize downlink of critical features [1], draft summary maps of visited terrain [2], or identify targets of opportunity for followup measurements [3]. We describe a generic approach to classify geologic surfaces for autonomous science operations, suitable for parallelized implementations in FPGA hardware. We map these surfaces with texture channels- distinctive numerical signatures that differentiate properties such as roughness, pavement coatings, regolith characteristics, sedimentary fabrics and differential outcrop weathering. This work describes our basic image analysis approach and reports an initial performance evaluation using surface images from the Mars Exploration Rovers. Future work will incorporate these methods into camera hardware for real-time processing.

Serum and Prostate Cancer Tissue Signatures of ERG Rearrangement Derived from Quantitative Analysis of the PTEN Conditional Knockout Mouse Proteome.

Niels J. Rupp, Igor Cima, Ralph Schiess, Peter J. Schüffler, Thomas J. Fuchs, Niklaus Frankhauser, Martin Kälin, Silke Gillessen, Ruedi Aebersold, Wilhelm Krek, Mark A. Rubin, Holger Moch and Peter J. Wild.

Symposium of the German Society for Pathology, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{rupp_serum_2012,
    title = {Serum and {Prostate} {Cancer} {Tissue} {Signatures} of {ERG} {Rearrangement} {Derived} from {Quantitative} {Analysis} of the {PTEN} {Conditional} {Knockout} {Mouse} {Proteome}},
    abstract = {Aims: Applying a systems biology approach to assess serum and tissue signatures of ERG rearrangement in patients with prostate cancer with the goal of determining downstream molecular targets for gene fusion cancers.
    Methods: Prostate tissue from a conditional PTEN knockout mouse model of prostate cancer was investigated, using selective enrichment of N-glycopeptides and mass spectrometry-based label-free quantification. Mouse tissue signatures were validated in sera and tissue of mice (n=12) and humans (n=105) by selected reaction monitoring (SRM), ELISA, and immunohistochemistry. ERG rearrangement status was assessed using fluorescence in situ hybridization (FISH) on two independent tissue microarray based prostatectomy cohorts (n=41, n=348). A Random forest model was trained and validated to identify serum and tissue signatures of ERG rearrangement.
    Results: A comprehensive PTEN dependent protein catalogue representing over 700 glycoproteins was established. TMPRSS2-ERG gene fusions occured in 41\% (17/41) of analyzable prostate cancers of the training cohort. ERG dependent serum signatures could be found. Serum signatures were tested on prostate cancer tissue microarrays using immunohistochemistry (test cohort).
    Conclusions: This is the first study to demonstrate a proteomic signature of ERG rearrangement prostate cancer. Given the challenges of directly targeting ETS transcription factors, this study has potential clinical implications providing important insights into future targetable downstream pathways.},
    booktitle = {Symposium of the {German} {Society} for {Pathology}},
    author = {Rupp, Niels J. and Cima, Igor and Schiess, Ralph and Sch\"uffler, Peter J. and Fuchs, Thomas J. and Frankhauser, Niklaus and K\"alin, Martin and Gillessen, Silke and Aebersold, Ruedi and Krek, Wilhelm and Rubin, Mark A. and Moch, Holger and Wild, Peter J.},
    year = {2012},
}

Download Endnote/RIS citation

TY - CONF
TI - Serum and Prostate Cancer Tissue Signatures of ERG Rearrangement Derived from Quantitative Analysis of the PTEN Conditional Knockout Mouse Proteome
AU - Rupp, Niels J.
AU - Cima, Igor
AU - Schiess, Ralph
AU - Schüffler, Peter J.
AU - Fuchs, Thomas J.
AU - Frankhauser, Niklaus
AU - Kälin, Martin
AU - Gillessen, Silke
AU - Aebersold, Ruedi
AU - Krek, Wilhelm
AU - Rubin, Mark A.
AU - Moch, Holger
AU - Wild, Peter J.
AB - Aims: Applying a systems biology approach to assess serum and tissue signatures of ERG rearrangement in patients with prostate cancer with the goal of determining downstream molecular targets for gene fusion cancers.
Methods: Prostate tissue from a conditional PTEN knockout mouse model of prostate cancer was investigated, using selective enrichment of N-glycopeptides and mass spectrometry-based label-free quantification. Mouse tissue signatures were validated in sera and tissue of mice (n=12) and humans (n=105) by selected reaction monitoring (SRM), ELISA, and immunohistochemistry. ERG rearrangement status was assessed using fluorescence in situ hybridization (FISH) on two independent tissue microarray based prostatectomy cohorts (n=41, n=348). A Random forest model was trained and validated to identify serum and tissue signatures of ERG rearrangement.
Results: A comprehensive PTEN dependent protein catalogue representing over 700 glycoproteins was established. TMPRSS2-ERG gene fusions occured in 41% (17/41) of analyzable prostate cancers of the training cohort. ERG dependent serum signatures could be found. Serum signatures were tested on prostate cancer tissue microarrays using immunohistochemistry (test cohort).
Conclusions: This is the first study to demonstrate a proteomic signature of ERG rearrangement prostate cancer. Given the challenges of directly targeting ETS transcription factors, this study has potential clinical implications providing important insights into future targetable downstream pathways.
C3 - Symposium of the German Society for Pathology
DA - 2012///
PY - 2012
ER -

Aims: Applying a systems biology approach to assess serum and tissue signatures of ERG rearrangement in patients with prostate cancer with the goal of determining downstream molecular targets for gene fusion cancers. Methods: Prostate tissue from a conditional PTEN knockout mouse model of prostate cancer was investigated, using selective enrichment of N-glycopeptides and mass spectrometry-based label-free quantification. Mouse tissue signatures were validated in sera and tissue of mice (n=12) and humans (n=105) by selected reaction monitoring (SRM), ELISA, and immunohistochemistry. ERG rearrangement status was assessed using fluorescence in situ hybridization (FISH) on two independent tissue microarray based prostatectomy cohorts (n=41, n=348). A Random forest model was trained and validated to identify serum and tissue signatures of ERG rearrangement. Results: A comprehensive PTEN dependent protein catalogue representing over 700 glycoproteins was established. TMPRSS2-ERG gene fusions occured in 41% (17/41) of analyzable prostate cancers of the training cohort. ERG dependent serum signatures could be found. Serum signatures were tested on prostate cancer tissue microarrays using immunohistochemistry (test cohort). Conclusions: This is the first study to demonstrate a proteomic signature of ERG rearrangement prostate cancer. Given the challenges of directly targeting ETS transcription factors, this study has potential clinical implications providing important insights into future targetable downstream pathways.

A High-Throughput Metabolomics Method to Predict High Concentration Cytotoxicity of Drugs from Low Concentration Profiles.

Stephanie Heux, Thomas J. Fuchs, Joachim Buhmann, Nicola Zamboni and Uwe Sauer.

Metabolomics, vol. 8, 3, p. 433-443, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{heux_high-throughput_2012,
    title = {A {High}-{Throughput} {Metabolomics} {Method} to {Predict} {High} {Concentration} {Cytotoxicity} of {Drugs} from {Low} {Concentration} {Profiles}},
    volume = {8},
    issn = {1573-3882},
    url = {http://dx.doi.org/10.1007/s11306-011-0386-0},
    doi = {10.1007/s11306-011-0386-0},
    number = {3},
    journal = {Metabolomics},
    author = {Heux, Stephanie and Fuchs, Thomas J. and Buhmann, Joachim and Zamboni, Nicola and Sauer, Uwe},
    year = {2012},
    keywords = {Biomedical and Life Sciences},
    pages = {433--443},
}

Download Endnote/RIS citation

TY - JOUR
TI - A High-Throughput Metabolomics Method to Predict High Concentration Cytotoxicity of Drugs from Low Concentration Profiles
AU - Heux, Stephanie
AU - Fuchs, Thomas J.
AU - Buhmann, Joachim
AU - Zamboni, Nicola
AU - Sauer, Uwe
T2 - Metabolomics
DA - 2012///
PY - 2012
DO - 10.1007/s11306-011-0386-0
VL - 8
IS - 3
SP - 433
EP - 443
SN - 1573-3882
UR - http://dx.doi.org/10.1007/s11306-011-0386-0
KW - Biomedical and Life Sciences
ER -

TMARKER: A User-Friendly Open-Source Assistance for Tma Grading and Cell Counting.

Peter J. Schueffler, Thomas J. Fuchs, Cheng S. Ong, Peter Wild and Joachim M. Buhmann.

Histopathology Image Analysis (HIMA) Workshop at the 15th International Conference on Medical Image Computing and Computer Assisted Intervention MICCAI, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schueffler_tmarker:_2012,
    title = {{TMARKER}: {A} {User}-{Friendly} {Open}-{Source} {Assistance} for {Tma} {Grading} and {Cell} {Counting}},
    booktitle = {Histopathology {Image} {Analysis} ({HIMA}) {Workshop} at the 15th {International} {Conference} on {Medical} {Image} {Computing} and {Computer} {Assisted} {Intervention} {MICCAI}},
    author = {Schueffler, Peter J. and Fuchs, Thomas J. and Ong, Cheng S. and Wild, Peter and Buhmann, Joachim M.},
    year = {2012},
}

Download Endnote/RIS citation

TY - CONF
TI - TMARKER: A User-Friendly Open-Source Assistance for Tma Grading and Cell Counting
AU - Schueffler, Peter J.
AU - Fuchs, Thomas J.
AU - Ong, Cheng S.
AU - Wild, Peter
AU - Buhmann, Joachim M.
C3 - Histopathology Image Analysis (HIMA) Workshop at the 15th International Conference on Medical Image Computing and Computer Assisted Intervention MICCAI
DA - 2012///
PY - 2012
ER -

A Seven-Marker Signature and Clinical Outcome in Malignant Melanoma: A Large-Scale Tissue-Microarray Study with Two Independent Patient Cohorts.

Stefanie Meyer, Thomas J. Fuchs, Anja K. Bosserhoff, Ferdinand Hofstädter, Armin Pauer, Volker Roth, Joachim M. Buhmann, Ingrid Moll, Nikos Anagnostou, Johanna M. Brandner, Kristian Ikenberg, Holger Moch, Michael Landthaler, Thomas Vogt and Peter J. Wild.

PLoS ONE, vol. 7, 6, p. e38222, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{meyer_seven-marker_2012,
    title = {A {Seven}-{Marker} {Signature} and {Clinical} {Outcome} in {Malignant} {Melanoma}: {A} {Large}-{Scale} {Tissue}-{Microarray} {Study} with {Two} {Independent} {Patient} {Cohorts}},
    volume = {7},
    shorttitle = {A {Seven}-{Marker} {Signature} and {Clinical} {Outcome} in {Malignant} {Melanoma}},
    url = {http://dx.doi.org/10.1371/journal.pone.0038222},
    doi = {10.1371/journal.pone.0038222},
    abstract = {BackgroundCurrent staging methods such as tumor thickness, ulceration and invasion of the sentinel node are known to be prognostic parameters in patients with malignant melanoma (MM). However, predictive molecular marker profiles for risk stratification and therapy optimization are not yet available for routine clinical assessment.Methods and FindingsUsing tissue microarrays, we retrospectively analyzed samples from 364 patients with primary MM. We investigated a panel of 70 immunohistochemical (IHC) antibodies for cell cycle, apoptosis, DNA mismatch repair, differentiation, proliferation, cell adhesion, signaling and metabolism. A marker selection procedure based on univariate Cox regression and multiple testing correction was employed to correlate the IHC expression data with the clinical follow-up (overall and recurrence-free survival). The model was thoroughly evaluated with two different cross validation experiments, a permutation test and a multivariate Cox regression analysis. In addition, the predictive power of the identified marker signature was validated on a second independent external test cohort (n = 225). A signature of seven biomarkers (Bax, Bcl-X, PTEN, COX-2, loss of β-Catenin, loss of MTAP, and presence of CD20 positive B-lymphocytes) was found to be an independent negative predictor for overall and recurrence-free survival in patients with MM. The seven-marker signature could also predict a high risk of disease recurrence in patients with localized primary MM stage pT1-2 (tumor thickness ≤2.00 mm). In particular, three of these markers (MTAP, COX-2, Bcl-X) were shown to offer direct therapeutic implications.ConclusionsThe seven-marker signature might serve as a prognostic tool enabling physicians to selectively triage, at the time of diagnosis, the subset of high recurrence risk stage I–II patients for adjuvant therapy. Selective treatment of those patients that are more likely to develop distant metastatic disease could potentially lower the burden of untreatable metastatic melanoma and revolutionize the therapeutic management of MM.},
    number = {6},
    urldate = {2016-01-03},
    journal = {PLoS ONE},
    author = {Meyer, Stefanie and Fuchs, Thomas J. and Bosserhoff, Anja K. and Hofst\"adter, Ferdinand and Pauer, Armin and Roth, Volker and Buhmann, Joachim M. and Moll, Ingrid and Anagnostou, Nikos and Brandner, Johanna M. and Ikenberg, Kristian and Moch, Holger and Landthaler, Michael and Vogt, Thomas and Wild, Peter J.},
    year = {2012},
    pages = {e38222},
}

Download Endnote/RIS citation

TY - JOUR
TI - A Seven-Marker Signature and Clinical Outcome in Malignant Melanoma: A Large-Scale Tissue-Microarray Study with Two Independent Patient Cohorts
AU - Meyer, Stefanie
AU - Fuchs, Thomas J.
AU - Bosserhoff, Anja K.
AU - Hofstädter, Ferdinand
AU - Pauer, Armin
AU - Roth, Volker
AU - Buhmann, Joachim M.
AU - Moll, Ingrid
AU - Anagnostou, Nikos
AU - Brandner, Johanna M.
AU - Ikenberg, Kristian
AU - Moch, Holger
AU - Landthaler, Michael
AU - Vogt, Thomas
AU - Wild, Peter J.
T2 - PLoS ONE
AB - BackgroundCurrent staging methods such as tumor thickness, ulceration and invasion of the sentinel node are known to be prognostic parameters in patients with malignant melanoma (MM). However, predictive molecular marker profiles for risk stratification and therapy optimization are not yet available for routine clinical assessment.Methods and FindingsUsing tissue microarrays, we retrospectively analyzed samples from 364 patients with primary MM. We investigated a panel of 70 immunohistochemical (IHC) antibodies for cell cycle, apoptosis, DNA mismatch repair, differentiation, proliferation, cell adhesion, signaling and metabolism. A marker selection procedure based on univariate Cox regression and multiple testing correction was employed to correlate the IHC expression data with the clinical follow-up (overall and recurrence-free survival). The model was thoroughly evaluated with two different cross validation experiments, a permutation test and a multivariate Cox regression analysis. In addition, the predictive power of the identified marker signature was validated on a second independent external test cohort (n = 225). A signature of seven biomarkers (Bax, Bcl-X, PTEN, COX-2, loss of β-Catenin, loss of MTAP, and presence of CD20 positive B-lymphocytes) was found to be an independent negative predictor for overall and recurrence-free survival in patients with MM. The seven-marker signature could also predict a high risk of disease recurrence in patients with localized primary MM stage pT1-2 (tumor thickness ≤2.00 mm). In particular, three of these markers (MTAP, COX-2, Bcl-X) were shown to offer direct therapeutic implications.ConclusionsThe seven-marker signature might serve as a prognostic tool enabling physicians to selectively triage, at the time of diagnosis, the subset of high recurrence risk stage I–II patients for adjuvant therapy. Selective treatment of those patients that are more likely to develop distant metastatic disease could potentially lower the burden of untreatable metastatic melanoma and revolutionize the therapeutic management of MM.
DA - 2012///
PY - 2012
DO - 10.1371/journal.pone.0038222
DP - PLoS Journals
VL - 7
IS - 6
SP - e38222
J2 - PLoS ONE
ST - A Seven-Marker Signature and Clinical Outcome in Malignant Melanoma
UR - http://dx.doi.org/10.1371/journal.pone.0038222
Y2 - 2016/01/03/15:45:02
ER -

BackgroundCurrent staging methods such as tumor thickness, ulceration and invasion of the sentinel node are known to be prognostic parameters in patients with malignant melanoma (MM). However, predictive molecular marker profiles for risk stratification and therapy optimization are not yet available for routine clinical assessment.Methods and FindingsUsing tissue microarrays, we retrospectively analyzed samples from 364 patients with primary MM. We investigated a panel of 70 immunohistochemical (IHC) antibodies for cell cycle, apoptosis, DNA mismatch repair, differentiation, proliferation, cell adhesion, signaling and metabolism. A marker selection procedure based on univariate Cox regression and multiple testing correction was employed to correlate the IHC expression data with the clinical follow-up (overall and recurrence-free survival). The model was thoroughly evaluated with two different cross validation experiments, a permutation test and a multivariate Cox regression analysis. In addition, the predictive power of the identified marker signature was validated on a second independent external test cohort (n = 225). A signature of seven biomarkers (Bax, Bcl-X, PTEN, COX-2, loss of β-Catenin, loss of MTAP, and presence of CD20 positive B-lymphocytes) was found to be an independent negative predictor for overall and recurrence-free survival in patients with MM. The seven-marker signature could also predict a high risk of disease recurrence in patients with localized primary MM stage pT1-2 (tumor thickness ≤2.00 mm). In particular, three of these markers (MTAP, COX-2, Bcl-X) were shown to offer direct therapeutic implications.ConclusionsThe seven-marker signature might serve as a prognostic tool enabling physicians to selectively triage, at the time of diagnosis, the subset of high recurrence risk stage I–II patients for adjuvant therapy. Selective treatment of those patients that are more likely to develop distant metastatic disease could potentially lower the burden of untreatable metastatic melanoma and revolutionize the therapeutic management of MM.

Combined Shape, Appearance and Silhouette for Simultaneous Manipulator and Object Tracking.

Paul Hebert, Nicolas Hudson, Jeremy Ma, Thomas Howard, Thomas J. Fuchs, Max Bajracharya and Joel Burdick.

IEEE International Conference on Robotics and Automation (ICRA), p. 2405-2412, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{hebert_combined_2012,
    title = {Combined {Shape}, {Appearance} and {Silhouette} for {Simultaneous} {Manipulator} and {Object} {Tracking}},
    url = {http://robotics.caltech.edu/wiki/images/d/d3/HebertICRA12.pdf},
    doi = {10.1109/ICRA.2012.6225084},
    booktitle = {{IEEE} {International} {Conference} on {Robotics} and {Automation} ({ICRA})},
    author = {Hebert, Paul and Hudson, Nicolas and Ma, Jeremy and Howard, Thomas and Fuchs, Thomas J. and Bajracharya, Max and Burdick, Joel},
    year = {2012},
    pages = {2405--2412},
}

Download Endnote/RIS citation

TY - CONF
TI - Combined Shape, Appearance and Silhouette for Simultaneous Manipulator and Object Tracking
AU - Hebert, Paul
AU - Hudson, Nicolas
AU - Ma, Jeremy
AU - Howard, Thomas
AU - Fuchs, Thomas J.
AU - Bajracharya, Max
AU - Burdick, Joel
C3 - IEEE International Conference on Robotics and Automation (ICRA)
DA - 2012///
PY - 2012
DO - 10.1109/ICRA.2012.6225084
SP - 2405
EP - 2412
UR - http://robotics.caltech.edu/wiki/images/d/d3/HebertICRA12.pdf
ER -

TextureCam: Autonomous Image Analysis for Astrobiology Survey.

David R. Thompson, Abigail Allwood, Dmitriy Bekker, Nathalie A. Cabrol, Tara Estlin, Thomas J. Fuchs and Kiri L. Wagstaff.

43rd Lunar and Planetary Science Conference, vol. 43, p. 1659, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{thompson_texturecam:_2012,
    title = {{TextureCam}: {Autonomous} {Image} {Analysis} for {Astrobiology} {Survey}},
    volume = {43},
    url = {http://adsabs.harvard.edu/abs/2012LPI....43.1659T},
    booktitle = {43rd {Lunar} and {Planetary} {Science} {Conference}},
    author = {Thompson, David R. and Allwood, Abigail and Bekker, Dmitriy and Cabrol, Nathalie A. and Estlin, Tara and Fuchs, Thomas J. and Wagstaff, Kiri L.},
    year = {2012},
    pages = {1659},
}

Download Endnote/RIS citation

TY - CONF
TI - TextureCam: Autonomous Image Analysis for Astrobiology Survey
AU - Thompson, David R.
AU - Allwood, Abigail
AU - Bekker, Dmitriy
AU - Cabrol, Nathalie A.
AU - Estlin, Tara
AU - Fuchs, Thomas J.
AU - Wagstaff, Kiri L.
C3 - 43rd Lunar and Planetary Science Conference
DA - 2012///
PY - 2012
VL - 43
SP - 1659
UR - http://adsabs.harvard.edu/abs/2012LPI....43.1659T
ER -

Cancer Genetics-Guided Discovery of Serum Biomarker Signatures for Diagnosis and Prognosis of Prostate Cancer.

Igor Cima, Ralph Schiess, Peter Wild, Martin Kaelin, Peter Schueffler, Vinzenz Lange, Paola Picotti, Reto Ossola, Arnoud Templeton, Olga Schubert, Thomas J. Fuchs, Thomas Leippold, Stephen Wyler, Jens Zehetner, Wolfram Jochum, Joachim Buhmann, Thomas Cerny, Holger Moch, Silke Gillessen, Ruedi Aebersold and Wilhelm Krek.

PNAS: Proceedings of the National Academy of Sciences, vol. 108, 8, p. 3342-3347, 2011

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{cima_cancer_2011,
    title = {Cancer {Genetics}-{Guided} {Discovery} of {Serum} {Biomarker} {Signatures} for {Diagnosis} and {Prognosis} of {Prostate} {Cancer}},
    volume = {108},
    url = {http://www.pnas.org/content/108/8/3342.abstract},
    doi = {10.1073/pnas.1013699108},
    abstract = {A key barrier to the realization of personalized medicine for cancer is the identification of biomarkers. Here we describe a two-stage strategy for the discovery of serum biomarker signatures corresponding to specific cancer-causing mutations and its application to prostate cancer (PCa) in the context of the commonly occurring phosphatase and tensin homolog (PTEN) tumor-suppressor gene inactivation. In the first stage of our approach, we identified 775 N-linked glycoproteins from sera and prostate tissue of wild-type and Pten-null mice. Using label-free quantitative proteomics, we showed that Pten inactivation leads to measurable perturbations in the murine prostate and serum glycoproteome. Following bioinformatic prioritization, in a second stage we applied targeted proteomics to detect and quantify 39 human ortholog candidate biomarkers in the sera of PCa patients and control individuals. The resulting proteomic profiles were analyzed by machine learning to build predictive regression models for tissue PTEN status and diagnosis and grading of PCa. Our approach suggests a general path to rational cancer biomarker discovery and initial validation guided by cancer genetics and based on the integration of experimental mouse models, proteomics-based technologies, and computational modeling.},
    number = {8},
    journal = {PNAS: Proceedings of the National Academy of Sciences},
    author = {Cima, Igor and Schiess, Ralph and Wild, Peter and Kaelin, Martin and Schueffler, Peter and Lange, Vinzenz and Picotti, Paola and Ossola, Reto and Templeton, Arnoud and Schubert, Olga and Fuchs, Thomas J. and Leippold, Thomas and Wyler, Stephen and Zehetner, Jens and Jochum, Wolfram and Buhmann, Joachim and Cerny, Thomas and Moch, Holger and Gillessen, Silke and Aebersold, Ruedi and Krek, Wilhelm},
    year = {2011},
    pages = {3342--3347},
}

Download Endnote/RIS citation

TY - JOUR
TI - Cancer Genetics-Guided Discovery of Serum Biomarker Signatures for Diagnosis and Prognosis of Prostate Cancer
AU - Cima, Igor
AU - Schiess, Ralph
AU - Wild, Peter
AU - Kaelin, Martin
AU - Schueffler, Peter
AU - Lange, Vinzenz
AU - Picotti, Paola
AU - Ossola, Reto
AU - Templeton, Arnoud
AU - Schubert, Olga
AU - Fuchs, Thomas J.
AU - Leippold, Thomas
AU - Wyler, Stephen
AU - Zehetner, Jens
AU - Jochum, Wolfram
AU - Buhmann, Joachim
AU - Cerny, Thomas
AU - Moch, Holger
AU - Gillessen, Silke
AU - Aebersold, Ruedi
AU - Krek, Wilhelm
T2 - PNAS: Proceedings of the National Academy of Sciences
AB - A key barrier to the realization of personalized medicine for cancer is the identification of biomarkers. Here we describe a two-stage strategy for the discovery of serum biomarker signatures corresponding to specific cancer-causing mutations and its application to prostate cancer (PCa) in the context of the commonly occurring phosphatase and tensin homolog (PTEN) tumor-suppressor gene inactivation. In the first stage of our approach, we identified 775 N-linked glycoproteins from sera and prostate tissue of wild-type and Pten-null mice. Using label-free quantitative proteomics, we showed that Pten inactivation leads to measurable perturbations in the murine prostate and serum glycoproteome. Following bioinformatic prioritization, in a second stage we applied targeted proteomics to detect and quantify 39 human ortholog candidate biomarkers in the sera of PCa patients and control individuals. The resulting proteomic profiles were analyzed by machine learning to build predictive regression models for tissue PTEN status and diagnosis and grading of PCa. Our approach suggests a general path to rational cancer biomarker discovery and initial validation guided by cancer genetics and based on the integration of experimental mouse models, proteomics-based technologies, and computational modeling.
DA - 2011///
PY - 2011
DO - 10.1073/pnas.1013699108
VL - 108
IS - 8
SP - 3342
EP - 3347
UR - http://www.pnas.org/content/108/8/3342.abstract
ER -

A key barrier to the realization of personalized medicine for cancer is the identification of biomarkers. Here we describe a two-stage strategy for the discovery of serum biomarker signatures corresponding to specific cancer-causing mutations and its application to prostate cancer (PCa) in the context of the commonly occurring phosphatase and tensin homolog (PTEN) tumor-suppressor gene inactivation. In the first stage of our approach, we identified 775 N-linked glycoproteins from sera and prostate tissue of wild-type and Pten-null mice. Using label-free quantitative proteomics, we showed that Pten inactivation leads to measurable perturbations in the murine prostate and serum glycoproteome. Following bioinformatic prioritization, in a second stage we applied targeted proteomics to detect and quantify 39 human ortholog candidate biomarkers in the sera of PCa patients and control individuals. The resulting proteomic profiles were analyzed by machine learning to build predictive regression models for tissue PTEN status and diagnosis and grading of PCa. Our approach suggests a general path to rational cancer biomarker discovery and initial validation guided by cancer genetics and based on the integration of experimental mouse models, proteomics-based technologies, and computational modeling.

Aerosols Transmit Prions to Immunocompetent and Immunodeficient Mice.

Johannes Haybaeck, Mathias Heikenwalder, Britta Klevenz, Petra Schwarz, Ilan Margalith, Claire Bridel, Kirsten Mertz, Elizabeta Zirdum, Benjamin Petsch, Thomas J. Fuchs, Lothar Stitz and Adriano Aguzzi.

PLoS Pathog, vol. 7, 1, p. e1001257, 2011

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{haybaeck_aerosols_2011,
    title = {Aerosols {Transmit} {Prions} to {Immunocompetent} and {Immunodeficient} {Mice}},
    volume = {7},
    url = {http://dx.doi.org/10.1371/journal.ppat.1001257},
    doi = {10.1371/journal.ppat.1001257},
    abstract = {Author Summary
    Prions, which are the cause of fatal neurodegenerative disorders termed transmissible spongiform encephalopathies (TSEs), can be experimentally or naturally transmitted via prion-contaminated food, blood, milk, saliva, feces and urine. Here we demonstrate that prions can be transmitted through aerosols in mice. This also occurs in the absence of immune cells as demonstrated by experiments with mice lacking B-, T-, follicular dendritic cells (FDCs), lymphotoxin signaling or with complement-deficient mice. Therefore, a functionally intact immune system is not strictly needed for aerogenic prion infection. These results suggest that current biosafety guidelines applied in diagnostic and scientific laboratories ought to include prion aerosols as a potential vector for prion infection.},
    number = {1},
    urldate = {2016-01-03},
    journal = {PLoS Pathog},
    author = {Haybaeck, Johannes and Heikenwalder, Mathias and Klevenz, Britta and Schwarz, Petra and Margalith, Ilan and Bridel, Claire and Mertz, Kirsten and Zirdum, Elizabeta and Petsch, Benjamin and Fuchs, Thomas J. and Stitz, Lothar and Aguzzi, Adriano},
    year = {2011},
    pages = {e1001257},
}

Download Endnote/RIS citation

TY - JOUR
TI - Aerosols Transmit Prions to Immunocompetent and Immunodeficient Mice
AU - Haybaeck, Johannes
AU - Heikenwalder, Mathias
AU - Klevenz, Britta
AU - Schwarz, Petra
AU - Margalith, Ilan
AU - Bridel, Claire
AU - Mertz, Kirsten
AU - Zirdum, Elizabeta
AU - Petsch, Benjamin
AU - Fuchs, Thomas J.
AU - Stitz, Lothar
AU - Aguzzi, Adriano
T2 - PLoS Pathog
AB - Author Summary
Prions, which are the cause of fatal neurodegenerative disorders termed transmissible spongiform encephalopathies (TSEs), can be experimentally or naturally transmitted via prion-contaminated food, blood, milk, saliva, feces and urine. Here we demonstrate that prions can be transmitted through aerosols in mice. This also occurs in the absence of immune cells as demonstrated by experiments with mice lacking B-, T-, follicular dendritic cells (FDCs), lymphotoxin signaling or with complement-deficient mice. Therefore, a functionally intact immune system is not strictly needed for aerogenic prion infection. These results suggest that current biosafety guidelines applied in diagnostic and scientific laboratories ought to include prion aerosols as a potential vector for prion infection.
DA - 2011///
PY - 2011
DO - 10.1371/journal.ppat.1001257
DP - PLoS Journals
VL - 7
IS - 1
SP - e1001257
J2 - PLoS Pathog
UR - http://dx.doi.org/10.1371/journal.ppat.1001257
Y2 - 2016/01/03/16:08:54
ER -

Author Summary Prions, which are the cause of fatal neurodegenerative disorders termed transmissible spongiform encephalopathies (TSEs), can be experimentally or naturally transmitted via prion-contaminated food, blood, milk, saliva, feces and urine. Here we demonstrate that prions can be transmitted through aerosols in mice. This also occurs in the absence of immune cells as demonstrated by experiments with mice lacking B-, T-, follicular dendritic cells (FDCs), lymphotoxin signaling or with complement-deficient mice. Therefore, a functionally intact immune system is not strictly needed for aerogenic prion infection. These results suggest that current biosafety guidelines applied in diagnostic and scientific laboratories ought to include prion aerosols as a potential vector for prion infection.

Computational Pathology: Challenges and Promises for Tissue Analysis.

Thomas J. Fuchs and Joachim M. Buhmann.

Computerized Medical Imaging and Graphics, vol. 35, 7–8, p. 515-530, 2011

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{fuchs_computational_2011,
    title = {Computational {Pathology}: {Challenges} and {Promises} for {Tissue} {Analysis}},
    volume = {35},
    issn = {0895-6111},
    shorttitle = {Computational pathology},
    url = {http://www.sciencedirect.com/science/article/pii/S0895611111000383},
    doi = {10.1016/j.compmedimag.2011.02.006},
    abstract = {The histological assessment of human tissue has emerged as the key challenge for detection and treatment of cancer. A plethora of different data sources ranging from tissue microarray data to gene expression, proteomics or metabolomics data provide a detailed overview of the health status of a patient. Medical doctors need to assess these information sources and they rely on data driven automatic analysis tools. Methods for classification, grouping and segmentation of heterogeneous data sources as well as regression of noisy dependencies and estimation of survival probabilities enter the processing workflow of a pathology diagnosis system at various stages. This paper reports on state-of-the-art of the design and effectiveness of computational pathology workflows and it discusses future research directions in this emergent field of medical informatics and diagnostic machine learning.},
    number = {7–8},
    urldate = {2012-03-08},
    journal = {Computerized Medical Imaging and Graphics},
    author = {Fuchs, Thomas J. and Buhmann, Joachim M.},
    year = {2011},
    keywords = {Cancer research, Computational pathology, Medical imaging, Survival statistics, Whole, Whole slide imaging, imaging, machine learning, slide},
    pages = {515--530},
}

Download Endnote/RIS citation

TY - JOUR
TI - Computational Pathology: Challenges and Promises for Tissue Analysis
AU - Fuchs, Thomas J.
AU - Buhmann, Joachim M.
T2 - Computerized Medical Imaging and Graphics
AB - The histological assessment of human tissue has emerged as the key challenge for detection and treatment of cancer. A plethora of different data sources ranging from tissue microarray data to gene expression, proteomics or metabolomics data provide a detailed overview of the health status of a patient. Medical doctors need to assess these information sources and they rely on data driven automatic analysis tools. Methods for classification, grouping and segmentation of heterogeneous data sources as well as regression of noisy dependencies and estimation of survival probabilities enter the processing workflow of a pathology diagnosis system at various stages. This paper reports on state-of-the-art of the design and effectiveness of computational pathology workflows and it discusses future research directions in this emergent field of medical informatics and diagnostic machine learning.
DA - 2011///
PY - 2011
DO - 10.1016/j.compmedimag.2011.02.006
DP - ScienceDirect
VL - 35
IS - 7–8
SP - 515
EP - 530
SN - 0895-6111
ST - Computational pathology
UR - http://www.sciencedirect.com/science/article/pii/S0895611111000383
Y2 - 2012/03/08/15:27:48
KW - Cancer research
KW - Computational pathology
KW - Medical imaging
KW - Survival statistics
KW - Whole
KW - Whole slide imaging
KW - imaging
KW - machine learning
KW - slide
ER -

The histological assessment of human tissue has emerged as the key challenge for detection and treatment of cancer. A plethora of different data sources ranging from tissue microarray data to gene expression, proteomics or metabolomics data provide a detailed overview of the health status of a patient. Medical doctors need to assess these information sources and they rely on data driven automatic analysis tools. Methods for classification, grouping and segmentation of heterogeneous data sources as well as regression of noisy dependencies and estimation of survival probabilities enter the processing workflow of a pathology diagnosis system at various stages. This paper reports on state-of-the-art of the design and effectiveness of computational pathology workflows and it discusses future research directions in this emergent field of medical informatics and diagnostic machine learning.

Computational TMA Analysis and Cell Nucleus Classification of Renal Cell Carcinoma.

Peter J. Schueffler, Thomas J. Fuchs, Cheng Soon Ong, Volker Roth and Joachim M. Buhmann.

Proceedings of the 32nd DAGM conference on Pattern recognition, p. 202–211, Springer-Verlag, Berlin, Heidelberg, ISBN 3-642-15985-0 978-3-642-15985-5, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{schueffler_computational_2010,
    address = {Darmstadt, Germany},
    title = {Computational {TMA} {Analysis} and {Cell} {Nucleus} {Classification} of {Renal} {Cell} {Carcinoma}},
    isbn = {3-642-15985-0 978-3-642-15985-5},
    url = {http://portal.acm.org/citation.cfm?id=1926258.1926281},
    doi = {10.1007/978-3-642-15986-2_21},
    abstract = {We consider an automated processing pipeline for tissue micro array analysis (TMA) of renal cell carcinoma. It consists of several consecutive tasks, which can be mapped to machine learning challenges. We investigate three of these tasks, namely nuclei segmentation, nuclei classification and staining estimation. We argue for a holistic view of the processing pipeline, as it is not obvious whether performance improvements at individual steps improve overall accuracy. The experimental results show that classification accuracy, which is comparable to trained human experts, can be achieved by using support vector machines (SVM) with appropriate kernels. Furthermore, we provide evidence that the shape of cell nuclei increases the classification performance. Most importantly, these improvements in classification accuracy result in corresponding improvements for the medically relevant estimation of immunohistochemical staining.},
    booktitle = {Proceedings of the 32nd {DAGM} conference on {Pattern} recognition},
    publisher = {Springer-Verlag, Berlin, Heidelberg},
    author = {Schueffler, Peter J. and Fuchs, Thomas J. and Ong, Cheng Soon and Roth, Volker and Buhmann, Joachim M.},
    year = {2010},
    pages = {202--211},
}

Download Endnote/RIS citation

TY - CONF
TI - Computational TMA Analysis and Cell Nucleus Classification of Renal Cell Carcinoma
AU - Schueffler, Peter J.
AU - Fuchs, Thomas J.
AU - Ong, Cheng Soon
AU - Roth, Volker
AU - Buhmann, Joachim M.
AB - We consider an automated processing pipeline for tissue micro array analysis (TMA) of renal cell carcinoma. It consists of several consecutive tasks, which can be mapped to machine learning challenges. We investigate three of these tasks, namely nuclei segmentation, nuclei classification and staining estimation. We argue for a holistic view of the processing pipeline, as it is not obvious whether performance improvements at individual steps improve overall accuracy. The experimental results show that classification accuracy, which is comparable to trained human experts, can be achieved by using support vector machines (SVM) with appropriate kernels. Furthermore, we provide evidence that the shape of cell nuclei increases the classification performance. Most importantly, these improvements in classification accuracy result in corresponding improvements for the medically relevant estimation of immunohistochemical staining.
C1 - Darmstadt, Germany
C3 - Proceedings of the 32nd DAGM conference on Pattern recognition
DA - 2010///
PY - 2010
DO - 10.1007/978-3-642-15986-2_21
SP - 202
EP - 211
PB - Springer-Verlag, Berlin, Heidelberg
SN - 3-642-15985-0 978-3-642-15985-5
UR - http://portal.acm.org/citation.cfm?id=1926258.1926281
ER -

We consider an automated processing pipeline for tissue micro array analysis (TMA) of renal cell carcinoma. It consists of several consecutive tasks, which can be mapped to machine learning challenges. We investigate three of these tasks, namely nuclei segmentation, nuclei classification and staining estimation. We argue for a holistic view of the processing pipeline, as it is not obvious whether performance improvements at individual steps improve overall accuracy. The experimental results show that classification accuracy, which is comparable to trained human experts, can be achieved by using support vector machines (SVM) with appropriate kernels. Furthermore, we provide evidence that the shape of cell nuclei increases the classification performance. Most importantly, these improvements in classification accuracy result in corresponding improvements for the medically relevant estimation of immunohistochemical staining.

The Translation-invariant Wishart-Dirichlet Process for Clustering Distance Data.

Julia E. Vogt, Sandhya Prabhakaran, Thomas J. Fuchs and Volker Roth.

Proceedings of the 27th International Conference on Machine Learning, p. 1111-1118, ICML'10, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{vogt_translation-invariant_2010,
    series = {{ICML}'10},
    title = {The {Translation}-invariant {Wishart}-{Dirichlet} {Process} for {Clustering} {Distance} {Data}},
    url = {http://www.icml2010.org/papers/248.pdf},
    booktitle = {Proceedings of the 27th {International} {Conference} on {Machine} {Learning}},
    author = {Vogt, Julia E. and Prabhakaran, Sandhya and Fuchs, Thomas J. and Roth, Volker},
    year = {2010},
    pages = {1111--1118},
}

Download Endnote/RIS citation

TY - CONF
TI - The Translation-invariant Wishart-Dirichlet Process for Clustering Distance Data
AU - Vogt, Julia E.
AU - Prabhakaran, Sandhya
AU - Fuchs, Thomas J.
AU - Roth, Volker
T3 - ICML'10
C3 - Proceedings of the 27th International Conference on Machine Learning
DA - 2010///
PY - 2010
SP - 1111
EP - 1118
UR - http://www.icml2010.org/papers/248.pdf
ER -

Neuron Geometry Extraction by Perceptual Grouping in ssTEM Images.

Verena Kaynig, Thomas J. Fuchs and Joachim M. Buhmann.

IEEE Computer Society Conference on Computer Vision and Pattern Recognition, vol. 0, p. 2902-2909, IEEE Computer Society, ISBN 978-1-4244-6984-0, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{kaynig_neuron_2010,
    address = {Los Alamitos, CA, USA},
    title = {Neuron {Geometry} {Extraction} by {Perceptual} {Grouping} in {ssTEM} {Images}},
    volume = {0},
    isbn = {978-1-4244-6984-0},
    url = {http://www.computer.org/portal/web/csdl/doi/10.1109/CVPR.2010.5540029},
    doi = {http://doi.ieeecomputersociety.org/10.1109/CVPR.2010.5540029},
    booktitle = {{IEEE} {Computer} {Society} {Conference} on {Computer} {Vision} and {Pattern} {Recognition}},
    publisher = {IEEE Computer Society},
    author = {Kaynig, Verena and Fuchs, Thomas J. and Buhmann, Joachim M.},
    year = {2010},
    pages = {2902--2909},
}

Download Endnote/RIS citation

TY - CONF
TI - Neuron Geometry Extraction by Perceptual Grouping in ssTEM Images
AU - Kaynig, Verena
AU - Fuchs, Thomas J.
AU - Buhmann, Joachim M.
C1 - Los Alamitos, CA, USA
C3 - IEEE Computer Society Conference on Computer Vision and Pattern Recognition
DA - 2010///
PY - 2010
DO - http://doi.ieeecomputersociety.org/10.1109/CVPR.2010.5540029
VL - 0
SP - 2902
EP - 2909
PB - IEEE Computer Society
SN - 978-1-4244-6984-0
UR - http://www.computer.org/portal/web/csdl/doi/10.1109/CVPR.2010.5540029
ER -

TAK1 Suppresses a NEMO-Dependent but NF-kappaB-Independent Pathway to Liver Cancer.

Kira Bettermann, Mihael Vucur, Johannes Haybaeck, Christiane Koppe, Joern Janssen, Felix Heymann, Achim Weber, Ralf Weiskirchen, Christian Liedtke, Nikolaus Gassler, Michael Mueller, Rita de Vos, Monika Julia Wolf, Yannick Boege, Gitta Maria Seleznik, Nicolas Zeller, Daniel Erny, Thomas J. Fuchs, Stefan Zoller, Stefano Cairo, Marie-Annick Buendia, Marco Prinz, Shizuo Akira, Frank Tacke, Mathias Heikenwaelder, Christian Trautwein and Tom Luedde.

Cancer Cell, vol. 17, 5, p. 481–496, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{bettermann_tak1_2010,
    title = {{TAK1} {Suppresses} a {NEMO}-{Dependent} but {NF}-{kappaB}-{Independent} {Pathway} to {Liver} {Cancer}},
    volume = {17},
    url = {http://www.ncbi.nlm.nih.gov/pubmed/20478530},
    number = {5},
    journal = {Cancer Cell},
    author = {Bettermann, Kira and Vucur, Mihael and Haybaeck, Johannes and Koppe, Christiane and Janssen, Joern and Heymann, Felix and Weber, Achim and Weiskirchen, Ralf and Liedtke, Christian and Gassler, Nikolaus and Mueller, Michael and de Vos, Rita and Wolf, Monika Julia and Boege, Yannick and Seleznik, Gitta Maria and Zeller, Nicolas and Erny, Daniel and Fuchs, Thomas J. and Zoller, Stefan and Cairo, Stefano and Buendia, Marie-Annick and Prinz, Marco and Akira, Shizuo and Tacke, Frank and Heikenwaelder, Mathias and Trautwein, Christian and Luedde, Tom},
    year = {2010},
    pages = {481--496},
}

Download Endnote/RIS citation

TY - JOUR
TI - TAK1 Suppresses a NEMO-Dependent but NF-kappaB-Independent Pathway to Liver Cancer
AU - Bettermann, Kira
AU - Vucur, Mihael
AU - Haybaeck, Johannes
AU - Koppe, Christiane
AU - Janssen, Joern
AU - Heymann, Felix
AU - Weber, Achim
AU - Weiskirchen, Ralf
AU - Liedtke, Christian
AU - Gassler, Nikolaus
AU - Mueller, Michael
AU - de Vos, Rita
AU - Wolf, Monika Julia
AU - Boege, Yannick
AU - Seleznik, Gitta Maria
AU - Zeller, Nicolas
AU - Erny, Daniel
AU - Fuchs, Thomas J.
AU - Zoller, Stefan
AU - Cairo, Stefano
AU - Buendia, Marie-Annick
AU - Prinz, Marco
AU - Akira, Shizuo
AU - Tacke, Frank
AU - Heikenwaelder, Mathias
AU - Trautwein, Christian
AU - Luedde, Tom
T2 - Cancer Cell
DA - 2010///
PY - 2010
VL - 17
IS - 5
SP - 481
EP - 496
UR - http://www.ncbi.nlm.nih.gov/pubmed/20478530
ER -

Infinite Mixture-of-Experts Model for Sparse Survival Regression with Application to Breast Cancer.

Sudhir Raman, Thomas J. Fuchs, Peter J. Wild, Edgar Dahl, Joachim M. Buhmann and Volker Roth.

BMC Bioinformatics, vol. 11, 8, p. S8, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{raman_infinite_2010,
    title = {Infinite {Mixture}-of-{Experts} {Model} for {Sparse} {Survival} {Regression} with {Application} to {Breast} {Cancer}},
    volume = {11},
    issn = {1471-2105},
    url = {http://dx.doi.org/10.1186/1471-2105-11-S8-S8},
    doi = {10.1186/1471-2105-11-S8-S8},
    abstract = {We present an infinite mixture-of-experts model to find an unknown number of sub-groups within a given patient cohort based on survival analysis. The effect of patient features on survival is modeled using the Cox’s proportionality hazards model which yields a non-standard regression component. The model is able to find key explanatory factors (chosen from main effects and higher-order interactions) for each sub-group by enforcing sparsity on the regression coefficients via the Bayesian Group-Lasso.},
    number = {8},
    urldate = {2016-01-03},
    journal = {BMC Bioinformatics},
    author = {Raman, Sudhir and Fuchs, Thomas J. and Wild, Peter J. and Dahl, Edgar and Buhmann, Joachim M. and Roth, Volker},
    year = {2010},
    pages = {S8},
}

Download Endnote/RIS citation

TY - JOUR
TI - Infinite Mixture-of-Experts Model for Sparse Survival Regression with Application to Breast Cancer
AU - Raman, Sudhir
AU - Fuchs, Thomas J.
AU - Wild, Peter J.
AU - Dahl, Edgar
AU - Buhmann, Joachim M.
AU - Roth, Volker
T2 - BMC Bioinformatics
AB - We present an infinite mixture-of-experts model to find an unknown number of sub-groups within a given patient cohort based on survival analysis. The effect of patient features on survival is modeled using the Cox’s proportionality hazards model which yields a non-standard regression component. The model is able to find key explanatory factors (chosen from main effects and higher-order interactions) for each sub-group by enforcing sparsity on the regression coefficients via the Bayesian Group-Lasso.
DA - 2010///
PY - 2010
DO - 10.1186/1471-2105-11-S8-S8
DP - BioMed Central
VL - 11
IS - 8
SP - S8
J2 - BMC Bioinformatics
SN - 1471-2105
UR - http://dx.doi.org/10.1186/1471-2105-11-S8-S8
Y2 - 2016/01/03/16:02:59
ER -

We present an infinite mixture-of-experts model to find an unknown number of sub-groups within a given patient cohort based on survival analysis. The effect of patient features on survival is modeled using the Cox’s proportionality hazards model which yields a non-standard regression component. The model is able to find key explanatory factors (chosen from main effects and higher-order interactions) for each sub-group by enforcing sparsity on the regression coefficients via the Bayesian Group-Lasso.

Geometrical Consistent 3D Tracing of Neuronal Processes in ssTEM Data.

Verena Kaynig, Thomas J. Fuchs and Joachim M. Buhmann.

Proceedings of the 13th international conference on Medical image computing and computer-assisted intervention: Part II, p. 209–216, MICCAI'10, Springer-Verlag, Berlin, Heidelberg, ISBN 3-642-15744-0 978-3-642-15744-8, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{kaynig_geometrical_2010,
    address = {Beijing, China},
    series = {{MICCAI}'10},
    title = {Geometrical {Consistent} {3D} {Tracing} of {Neuronal} {Processes} in {ssTEM} {Data}},
    isbn = {3-642-15744-0 978-3-642-15744-8},
    url = {http://portal.acm.org/citation.cfm?id=1928047.1928075},
    booktitle = {Proceedings of the 13th international conference on {Medical} image computing and computer-assisted intervention: {Part} {II}},
    publisher = {Springer-Verlag, Berlin, Heidelberg},
    author = {Kaynig, Verena and Fuchs, Thomas J. and Buhmann, Joachim M.},
    year = {2010},
    pages = {209--216},
}

Download Endnote/RIS citation

TY - CONF
TI - Geometrical Consistent 3D Tracing of Neuronal Processes in ssTEM Data
AU - Kaynig, Verena
AU - Fuchs, Thomas J.
AU - Buhmann, Joachim M.
T3 - MICCAI'10
C1 - Beijing, China
C3 - Proceedings of the 13th international conference on Medical image computing and computer-assisted intervention: Part II
DA - 2010///
PY - 2010
SP - 209
EP - 216
PB - Springer-Verlag, Berlin, Heidelberg
SN - 3-642-15744-0 978-3-642-15744-8
UR - http://portal.acm.org/citation.cfm?id=1928047.1928075
ER -

Effect of Reader Experience on Variability, Evaluation Time and Accuracy of Coronary Plaque Detection with Computed Tomography Coronary Angiography.

Stefan Saur, Hatem Alkadhi, Paul Stolzmann, Stephan Baumueller, Sebastian Leschka, Hans Scheffel, Lotus Desbiolles, Thomas J. Fuchs, Gabor Szekely and Philippe Cattin.

European Radiology, vol. 20, 7, p. 1599-1606, 2010

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{saur_effect_2010,
    title = {Effect of {Reader} {Experience} on {Variability}, {Evaluation} {Time} and {Accuracy} of {Coronary} {Plaque} {Detection} with {Computed} {Tomography} {Coronary} {Angiography}},
    volume = {20},
    issn = {0938-7994},
    url = {http://dx.doi.org/10.1007/s00330-009-1709-7},
    doi = {10.1007/s00330-009-1709-7},
    number = {7},
    journal = {European Radiology},
    author = {Saur, Stefan and Alkadhi, Hatem and Stolzmann, Paul and Baumueller, Stephan and Leschka, Sebastian and Scheffel, Hans and Desbiolles, Lotus and Fuchs, Thomas J. and Szekely, Gabor and Cattin, Philippe},
    year = {2010},
    keywords = {Medicine},
    pages = {1599--1606},
}

Download Endnote/RIS citation

TY - JOUR
TI - Effect of Reader Experience on Variability, Evaluation Time and Accuracy of Coronary Plaque Detection with Computed Tomography Coronary Angiography
AU - Saur, Stefan
AU - Alkadhi, Hatem
AU - Stolzmann, Paul
AU - Baumueller, Stephan
AU - Leschka, Sebastian
AU - Scheffel, Hans
AU - Desbiolles, Lotus
AU - Fuchs, Thomas J.
AU - Szekely, Gabor
AU - Cattin, Philippe
T2 - European Radiology
DA - 2010///
PY - 2010
DO - 10.1007/s00330-009-1709-7
VL - 20
IS - 7
SP - 1599
EP - 1606
SN - 0938-7994
UR - http://dx.doi.org/10.1007/s00330-009-1709-7
KW - Medicine
ER -

Prognostic Relevance of Wnt-Inhibitory Factor-1 (WIF1) and Dickkopf-3 (DKK3) Promoter Methylation in Human Breast Cancer.

Jurgen Veeck, Peter Wild, Thomas J. Fuchs, Peter Schueffler, Arndt Hartmann, Ruth Knuchel and Edgar Dahl.

BMC Cancer, vol. 9, 1, p. 217, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{veeck_prognostic_2009,
    title = {Prognostic {Relevance} of {Wnt}-{Inhibitory} {Factor}-1 ({WIF1}) and {Dickkopf}-3 ({DKK3}) {Promoter} {Methylation} in {Human} {Breast} {Cancer}},
    volume = {9},
    issn = {1471-2407},
    url = {http://www.biomedcentral.com/1471-2407/9/217},
    doi = {10.1186/1471-2407-9-217},
    abstract = {Background
    Secreted Wnt signaling antagonists have recently been described as frequent targets of epigenetic inactivation in human tumor entities. Since gene silencing of certain Wnt antagonists was found to be correlated with adverse patient survival in cancer, we aimed at investigating a potential prognostic impact of the two Wnt antagonizing molecules WIF1 and DKK3 in breast cancer, which are frequently silenced by promoter methylation in this disease.
    Methods
    WIF1 and DKK3 promoter methylation were assessed by methylation-specific PCR with bisulfite-converted DNA from 19 normal breast tissues and 150 primary breast carcinomas. Promoter methylation was interpreted in a qualitative, binary fashion. Statistical evaluations included two-sided Fisher's exact tests, univariate log-rank tests of Kaplan-Meier curves as well as multivariate Cox regression analyses.
    Results
    WIF1 and DKK3 promoter methylation were detected in 63.3\% (95/150) and 61.3\% (92/150) of breast carcinoma samples, respectively. In normal breast tissues, WIF1 methylation was present in 0\% (0/19) and DKK3 methylation in 5.3\% (1/19) of samples. In breast carcinomas, WIF1 methylation was significantly associated with methylation of DKK3 (p = 0.009). Methylation of either gene was not associated with clinicopathological parameters, except for DKK3 methylation being associated with patient age (p = 0.007). In univariate analysis, WIF1 methylation was not associated with clinical patient outcome. In contrast, DKK3 methylation was a prognostic factor in patient overall survival (OS) and disease-free survival (DFS). Estimated OS rates after 10 years were 54\% for patients with DKK3 -methylated tumors, in contrast to patients without DKK3 methylation in the tumor, who had a favorable 97\% OS after 10 years (p {\textless} 0.001). Likewise, DFS at 10 years for patients harboring DKK3 methylation in the tumor was 58\%, compared with 78\% for patients with unmethylated DKK3 (p = 0.037). Multivariate analyses revealed that DKK3 methylation was an independent prognostic factor predicting poor OS (hazard ratio (HR): 14.4; 95\% confidence interval (CI): 1.9–111.6; p = 0.011), and short DFS (HR: 2.5; 95\% CI: 1.0–6.0; p = 0.047) in breast cancer.
    Conclusion
    Although the Wnt antagonist genes WIF1 and DKK3 show a very similar frequency of promoter methylation in human breast cancer, only DKK3 methylation proves as a novel prognostic marker potentially useful in the clinical management of this disease.},
    number = {1},
    journal = {BMC Cancer},
    author = {Veeck, Jurgen and Wild, Peter and Fuchs, Thomas J. and Schueffler, Peter and Hartmann, Arndt and Knuchel, Ruth and Dahl, Edgar},
    year = {2009},
    pages = {217},
}

Download Endnote/RIS citation

TY - JOUR
TI - Prognostic Relevance of Wnt-Inhibitory Factor-1 (WIF1) and Dickkopf-3 (DKK3) Promoter Methylation in Human Breast Cancer
AU - Veeck, Jurgen
AU - Wild, Peter
AU - Fuchs, Thomas J.
AU - Schueffler, Peter
AU - Hartmann, Arndt
AU - Knuchel, Ruth
AU - Dahl, Edgar
T2 - BMC Cancer
AB - Background
Secreted Wnt signaling antagonists have recently been described as frequent targets of epigenetic inactivation in human tumor entities. Since gene silencing of certain Wnt antagonists was found to be correlated with adverse patient survival in cancer, we aimed at investigating a potential prognostic impact of the two Wnt antagonizing molecules WIF1 and DKK3 in breast cancer, which are frequently silenced by promoter methylation in this disease.
Methods
WIF1 and DKK3 promoter methylation were assessed by methylation-specific PCR with bisulfite-converted DNA from 19 normal breast tissues and 150 primary breast carcinomas. Promoter methylation was interpreted in a qualitative, binary fashion. Statistical evaluations included two-sided Fisher's exact tests, univariate log-rank tests of Kaplan-Meier curves as well as multivariate Cox regression analyses.
Results
WIF1 and DKK3 promoter methylation were detected in 63.3% (95/150) and 61.3% (92/150) of breast carcinoma samples, respectively. In normal breast tissues, WIF1 methylation was present in 0% (0/19) and DKK3 methylation in 5.3% (1/19) of samples. In breast carcinomas, WIF1 methylation was significantly associated with methylation of DKK3 (p = 0.009). Methylation of either gene was not associated with clinicopathological parameters, except for DKK3 methylation being associated with patient age (p = 0.007). In univariate analysis, WIF1 methylation was not associated with clinical patient outcome. In contrast, DKK3 methylation was a prognostic factor in patient overall survival (OS) and disease-free survival (DFS). Estimated OS rates after 10 years were 54% for patients with DKK3 -methylated tumors, in contrast to patients without DKK3 methylation in the tumor, who had a favorable 97% OS after 10 years (p < 0.001). Likewise, DFS at 10 years for patients harboring DKK3 methylation in the tumor was 58%, compared with 78% for patients with unmethylated DKK3 (p = 0.037). Multivariate analyses revealed that DKK3 methylation was an independent prognostic factor predicting poor OS (hazard ratio (HR): 14.4; 95% confidence interval (CI): 1.9–111.6; p = 0.011), and short DFS (HR: 2.5; 95% CI: 1.0–6.0; p = 0.047) in breast cancer.
Conclusion
Although the Wnt antagonist genes WIF1 and DKK3 show a very similar frequency of promoter methylation in human breast cancer, only DKK3 methylation proves as a novel prognostic marker potentially useful in the clinical management of this disease.
DA - 2009///
PY - 2009
DO - 10.1186/1471-2407-9-217
VL - 9
IS - 1
SP - 217
SN - 1471-2407
UR - http://www.biomedcentral.com/1471-2407/9/217
ER -

Background Secreted Wnt signaling antagonists have recently been described as frequent targets of epigenetic inactivation in human tumor entities. Since gene silencing of certain Wnt antagonists was found to be correlated with adverse patient survival in cancer, we aimed at investigating a potential prognostic impact of the two Wnt antagonizing molecules WIF1 and DKK3 in breast cancer, which are frequently silenced by promoter methylation in this disease. Methods WIF1 and DKK3 promoter methylation were assessed by methylation-specific PCR with bisulfite-converted DNA from 19 normal breast tissues and 150 primary breast carcinomas. Promoter methylation was interpreted in a qualitative, binary fashion. Statistical evaluations included two-sided Fisher's exact tests, univariate log-rank tests of Kaplan-Meier curves as well as multivariate Cox regression analyses. Results WIF1 and DKK3 promoter methylation were detected in 63.3% (95/150) and 61.3% (92/150) of breast carcinoma samples, respectively. In normal breast tissues, WIF1 methylation was present in 0% (0/19) and DKK3 methylation in 5.3% (1/19) of samples. In breast carcinomas, WIF1 methylation was significantly associated with methylation of DKK3 (p = 0.009). Methylation of either gene was not associated with clinicopathological parameters, except for DKK3 methylation being associated with patient age (p = 0.007). In univariate analysis, WIF1 methylation was not associated with clinical patient outcome. In contrast, DKK3 methylation was a prognostic factor in patient overall survival (OS) and disease-free survival (DFS). Estimated OS rates after 10 years were 54% for patients with DKK3 -methylated tumors, in contrast to patients without DKK3 methylation in the tumor, who had a favorable 97% OS after 10 years (p < 0.001). Likewise, DFS at 10 years for patients harboring DKK3 methylation in the tumor was 58%, compared with 78% for patients with unmethylated DKK3 (p = 0.037). Multivariate analyses revealed that DKK3 methylation was an independent prognostic factor predicting poor OS (hazard ratio (HR): 14.4; 95% confidence interval (CI): 1.9–111.6; p = 0.011), and short DFS (HR: 2.5; 95% CI: 1.0–6.0; p = 0.047) in breast cancer. Conclusion Although the Wnt antagonist genes WIF1 and DKK3 show a very similar frequency of promoter methylation in human breast cancer, only DKK3 methylation proves as a novel prognostic marker potentially useful in the clinical management of this disease.

Graph-Based Pancreatic Islet Segmentation for Early Type 2 Diabetes Mellitus on Histopathological Tissue.

Xenofon E. Floros, Thomas J. Fuchs, Markus P. Rechsteiner, Giatgen Spinas, Holger Moch and Joachim M. Buhmann.

Medical Image Computing and Computer-Assisted Intervention MICCAI, p. 633-640, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{floros_graph-based_2009,
    title = {Graph-{Based} {Pancreatic} {Islet} {Segmentation} for {Early} {Type} 2 {Diabetes} {Mellitus} on {Histopathological} {Tissue}},
    url = {http://www.springerlink.com/index/2PG4628N77772L77.pdf},
    booktitle = {Medical {Image} {Computing} and {Computer}-{Assisted} {Intervention} {MICCAI}},
    author = {Floros, Xenofon E. and Fuchs, Thomas J. and Rechsteiner, Markus P. and Spinas, Giatgen and Moch, Holger and Buhmann, Joachim M.},
    year = {2009},
    pages = {633--640},
}

Download Endnote/RIS citation

TY - CONF
TI - Graph-Based Pancreatic Islet Segmentation for Early Type 2 Diabetes Mellitus on Histopathological Tissue
AU - Floros, Xenofon E.
AU - Fuchs, Thomas J.
AU - Rechsteiner, Markus P.
AU - Spinas, Giatgen
AU - Moch, Holger
AU - Buhmann, Joachim M.
C3 - Medical Image Computing and Computer-Assisted Intervention MICCAI
DA - 2009///
PY - 2009
SP - 633
EP - 640
UR - http://www.springerlink.com/index/2PG4628N77772L77.pdf
ER -

Detection of Urothelial Bladder Cancer Cells in Voided Urine Can Be Improved by a Combination of Cytology and Standardized Microsatellite Analysis.

Peter J. Wild, Thomas J. Fuchs, Robert Stoehr, Dieter Zimmermann, Simona Frigerio, Barbara Padberg, Inbal Steiner, Ellen C. Zwarthoff, Maximilian Burger, Stefan Denzinger, Ferdinand Hofstaedter, Glen Kristiansen, Thomas Hermanns, Hans-Helge Seifert, Maurizio Provenzano, Tullio Sulser, Volker Roth, Joachim M. Buhmann, Holger Moch and Arndt Hartmann.

Cancer Epidemiol Biomarkers Prev, vol. 18, 6, p. 1798-1806, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{wild_detection_2009,
    title = {Detection of {Urothelial} {Bladder} {Cancer} {Cells} in {Voided} {Urine} {Can} {Be} {Improved} by a {Combination} of {Cytology} and {Standardized} {Microsatellite} {Analysis}},
    volume = {18},
    url = {http://cebp.aacrjournals.org/cgi/content/abstract/18/6/1798},
    doi = {10.1158/1055-9965.EPI-09-0099},
    abstract = {Purpose: To evaluate molecular and immunohistochemical markers to develop a molecular grading of urothelial bladder cancer and to test these markers in voided urine samples.
    Experimental Design: 255 consecutive biopsies from primary bladder cancer patients were evaluated on a tissue microarray. The clinical parameters gender, age, adjacent carcinoma in situ, and multifocality were collected. UroVysion fluorescence in situ hybridization (FISH) was done. Expression of cytokeratin 20, MIB1, and TP53 was analyzed by immunohistochemistry. Fibroblast growth factor receptor 3 (FGFR3) status was studied by SNaPshot mutation detection. Results were correlated with clinical outcome by Cox regression analysis. To assess the predictive power of different predictor subsets to detect high grade and tumor invasion, logistic regression models were learned. Additionally, voided urine samples of 119 patients were investigated. After cytologic examination, urine samples were matched with their biopsies and analyzed for loss of heterozygosity (LOH), FGFR3 mutation, polysomy, and p16 deletion using UroVysion FISH. Receiver operator characteristic curves for various predictor subsets were plotted.
    Results: In biopsies, high grade and solid growth pattern were independent prognostic factors for overall survival. A model consisting of UroVysion FISH and FGFR3 status (FISH + FGFR3) predicted high grade significantly better compared with a recently proposed molecular grade (MIB1 + FGFR3). In voided urine, the combination of cytology with LOH analysis (CYTO + LOH) reached the highest diagnostic accuracy for the detection of bladder cancer cells and performed better than cytology alone (sensitivity of 88.2\% and specificity of 97.1\%).
    Conclusions: The combination of cytology with LOH analysis could reduce unpleasant cystoscopies for bladder cancer patients.},
    number = {6},
    journal = {Cancer Epidemiol Biomarkers Prev},
    author = {Wild, Peter J. and Fuchs, Thomas J. and Stoehr, Robert and Zimmermann, Dieter and Frigerio, Simona and Padberg, Barbara and Steiner, Inbal and Zwarthoff, Ellen C. and Burger, Maximilian and Denzinger, Stefan and Hofstaedter, Ferdinand and Kristiansen, Glen and Hermanns, Thomas and Seifert, Hans-Helge and Provenzano, Maurizio and Sulser, Tullio and Roth, Volker and Buhmann, Joachim M. and Moch, Holger and Hartmann, Arndt},
    year = {2009},
    pages = {1798--1806},
}

Download Endnote/RIS citation

TY - JOUR
TI - Detection of Urothelial Bladder Cancer Cells in Voided Urine Can Be Improved by a Combination of Cytology and Standardized Microsatellite Analysis
AU - Wild, Peter J.
AU - Fuchs, Thomas J.
AU - Stoehr, Robert
AU - Zimmermann, Dieter
AU - Frigerio, Simona
AU - Padberg, Barbara
AU - Steiner, Inbal
AU - Zwarthoff, Ellen C.
AU - Burger, Maximilian
AU - Denzinger, Stefan
AU - Hofstaedter, Ferdinand
AU - Kristiansen, Glen
AU - Hermanns, Thomas
AU - Seifert, Hans-Helge
AU - Provenzano, Maurizio
AU - Sulser, Tullio
AU - Roth, Volker
AU - Buhmann, Joachim M.
AU - Moch, Holger
AU - Hartmann, Arndt
T2 - Cancer Epidemiol Biomarkers Prev
AB - Purpose: To evaluate molecular and immunohistochemical markers to develop a molecular grading of urothelial bladder cancer and to test these markers in voided urine samples.
Experimental Design: 255 consecutive biopsies from primary bladder cancer patients were evaluated on a tissue microarray. The clinical parameters gender, age, adjacent carcinoma in situ, and multifocality were collected. UroVysion fluorescence in situ hybridization (FISH) was done. Expression of cytokeratin 20, MIB1, and TP53 was analyzed by immunohistochemistry. Fibroblast growth factor receptor 3 (FGFR3) status was studied by SNaPshot mutation detection. Results were correlated with clinical outcome by Cox regression analysis. To assess the predictive power of different predictor subsets to detect high grade and tumor invasion, logistic regression models were learned. Additionally, voided urine samples of 119 patients were investigated. After cytologic examination, urine samples were matched with their biopsies and analyzed for loss of heterozygosity (LOH), FGFR3 mutation, polysomy, and p16 deletion using UroVysion FISH. Receiver operator characteristic curves for various predictor subsets were plotted.
Results: In biopsies, high grade and solid growth pattern were independent prognostic factors for overall survival. A model consisting of UroVysion FISH and FGFR3 status (FISH + FGFR3) predicted high grade significantly better compared with a recently proposed molecular grade (MIB1 + FGFR3). In voided urine, the combination of cytology with LOH analysis (CYTO + LOH) reached the highest diagnostic accuracy for the detection of bladder cancer cells and performed better than cytology alone (sensitivity of 88.2% and specificity of 97.1%).
Conclusions: The combination of cytology with LOH analysis could reduce unpleasant cystoscopies for bladder cancer patients.
DA - 2009///
PY - 2009
DO - 10.1158/1055-9965.EPI-09-0099
VL - 18
IS - 6
SP - 1798
EP - 1806
UR - http://cebp.aacrjournals.org/cgi/content/abstract/18/6/1798
ER -

Purpose: To evaluate molecular and immunohistochemical markers to develop a molecular grading of urothelial bladder cancer and to test these markers in voided urine samples. Experimental Design: 255 consecutive biopsies from primary bladder cancer patients were evaluated on a tissue microarray. The clinical parameters gender, age, adjacent carcinoma in situ, and multifocality were collected. UroVysion fluorescence in situ hybridization (FISH) was done. Expression of cytokeratin 20, MIB1, and TP53 was analyzed by immunohistochemistry. Fibroblast growth factor receptor 3 (FGFR3) status was studied by SNaPshot mutation detection. Results were correlated with clinical outcome by Cox regression analysis. To assess the predictive power of different predictor subsets to detect high grade and tumor invasion, logistic regression models were learned. Additionally, voided urine samples of 119 patients were investigated. After cytologic examination, urine samples were matched with their biopsies and analyzed for loss of heterozygosity (LOH), FGFR3 mutation, polysomy, and p16 deletion using UroVysion FISH. Receiver operator characteristic curves for various predictor subsets were plotted. Results: In biopsies, high grade and solid growth pattern were independent prognostic factors for overall survival. A model consisting of UroVysion FISH and FGFR3 status (FISH + FGFR3) predicted high grade significantly better compared with a recently proposed molecular grade (MIB1 + FGFR3). In voided urine, the combination of cytology with LOH analysis (CYTO + LOH) reached the highest diagnostic accuracy for the detection of bladder cancer cells and performed better than cytology alone (sensitivity of 88.2% and specificity of 97.1%). Conclusions: The combination of cytology with LOH analysis could reduce unpleasant cystoscopies for bladder cancer patients.

Nuclear Detection of Y-Box Protein-1 (YB-1) Closely Associates with Progesterone Receptor Negativity and Is a Strong Adverse Survival Factor in Human Breast Cancer.

Edgar Dahl, Abdelaziz En-Nia, Frank Wiesmann, Renate Krings, Sonja Djudjaj, Elisabeth Breuer, Thomas J. Fuchs, Peter Wild, Arndt Hartmann, Sandra Dunn and Peter Mertens.

BMC Cancer, vol. 9, 1, p. 410, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{dahl_nuclear_2009,
    title = {Nuclear {Detection} of {Y}-{Box} {Protein}-1 ({YB}-1) {Closely} {Associates} with {Progesterone} {Receptor} {Negativity} and {Is} a {Strong} {Adverse} {Survival} {Factor} in {Human} {Breast} {Cancer}},
    volume = {9},
    issn = {1471-2407},
    url = {http://www.biomedcentral.com/1471-2407/9/410},
    doi = {10.1186/1471-2407-9-410},
    number = {1},
    journal = {BMC Cancer},
    author = {Dahl, Edgar and En-Nia, Abdelaziz and Wiesmann, Frank and Krings, Renate and Djudjaj, Sonja and Breuer, Elisabeth and Fuchs, Thomas J. and Wild, Peter and Hartmann, Arndt and Dunn, Sandra and Mertens, Peter},
    year = {2009},
    pages = {410},
}

Download Endnote/RIS citation

TY - JOUR
TI - Nuclear Detection of Y-Box Protein-1 (YB-1) Closely Associates with Progesterone Receptor Negativity and Is a Strong Adverse Survival Factor in Human Breast Cancer
AU - Dahl, Edgar
AU - En-Nia, Abdelaziz
AU - Wiesmann, Frank
AU - Krings, Renate
AU - Djudjaj, Sonja
AU - Breuer, Elisabeth
AU - Fuchs, Thomas J.
AU - Wild, Peter
AU - Hartmann, Arndt
AU - Dunn, Sandra
AU - Mertens, Peter
T2 - BMC Cancer
DA - 2009///
PY - 2009
DO - 10.1186/1471-2407-9-410
VL - 9
IS - 1
SP - 410
SN - 1471-2407
UR - http://www.biomedcentral.com/1471-2407/9/410
ER -

Prediction Rules for the Detection of Coronary Artery Plaques: Evidence from Cardiac CT.

Stefan C. Saur, Philippe C. Cattin, Lotus Desbiolles, Thomas J. Fuchs, Gabor Szekely and Hatem Alkadhi.

Investigative Radiology, vol. 44, 8, p. 483-490, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{saur_prediction_2009,
    title = {Prediction {Rules} for the {Detection} of {Coronary} {Artery} {Plaques}: {Evidence} from {Cardiac} {CT}},
    volume = {44},
    url = {http://journals.lww.com/investigativeradiology/Abstract/2009/08000/Prediction_Rules_for_the_Detection_of_Coronary.8.aspx},
    doi = {http://dx.doi.org/10.1097/RLI.0b013e3181a8afc4},
    abstract = {Objectives: To evaluate spatial plaque distribution patterns in coronary arteries based on computed tomography coronary angiography data sets and to express the learned patterns in prediction rules. An application is proposed to use these prediction rules for the detection of initially missed plaques.
    Material and Methods: Two hundred fifty two consecutive patients with chronic coronary artery disease underwent contrast-enhanced dual-source computed tomography coronary angiography for clinical indications. Coronary artery plaques were manually labeled on a 16-segment coronary model and their position (ie, segments and bifurcations) and composition (ie, calcified, mixed, or noncalcified) were noted. The frequent itemset mining algorithm was used to statistically search for plaque distribution patterns. The patterns were expressed as prediction rules: given plaques at certain locations as conditions, a prediction rule gave evidence—with a certain confidence value—for a plaque at another location within the coronary artery tree. Prediction rules with the highest confidence values were evaluated and described. Furthermore, to improve manual plaque detection, all prediction rules were applied on the patient data to search for segments with potentially missed plaques. These segments were then reviewed in a second, guided reading for the existence of plaques. The same number of segments was also determined by a weighted random approach to evaluate the quality of prediction resulting from frequent itemset mining.
    Results: In 200 of 252 (79.4\%) patients, at least one coronary plaque (range, 1–22 plaques) was found. In total 1229 plaques (990 calcified, 80.6\%; 227 mixed, 18.5\%; 12 noncalcified, 1\%) distributed, over 916 coronary segments and 507 vessels were manually labeled. Four plaque distribution patterns were identified: 20.6\% of the patients had no plaques at all; 31.7\% had plaques in the left coronary artery tree; 46.4\% had plaques both in left and right coronary arteries, whereas 1.2\% of the patients had plaques solely in the right coronary artery (RCA). General rules were found predicting plaques in the left anterior descending artery (LAD), given plaques in segments of the RCA or in the left main artery. Further general rules predicted plaques in the LAD, given plaques in the circumflex artery. In the guided review, the segment selection based on the prediction rules from frequent itemset mining performed significantly better (P {\textless} 0.001) than the weighted random approach by revealing 48 initially missed plaques.
    Conclusions: This study demonstrates spatial plaque distribution patterns in coronary arteries as determined with cardiac CT. Use of the frequent itemset mining algorithm yielded rules that predicted plaques at certain sites given plaques at other sites of the coronary artery tree. Use of these prediction rules improved the manual labeling of coronary plaques as initially missed plaques could be predicted with the guided review.},
    number = {8},
    journal = {Investigative Radiology},
    author = {Saur, Stefan C. and Cattin, Philippe C. and Desbiolles, Lotus and Fuchs, Thomas J. and Szekely, Gabor and Alkadhi, Hatem},
    year = {2009},
    pages = {483--490},
}

Download Endnote/RIS citation

TY - JOUR
TI - Prediction Rules for the Detection of Coronary Artery Plaques: Evidence from Cardiac CT
AU - Saur, Stefan C.
AU - Cattin, Philippe C.
AU - Desbiolles, Lotus
AU - Fuchs, Thomas J.
AU - Szekely, Gabor
AU - Alkadhi, Hatem
T2 - Investigative Radiology
AB - Objectives: To evaluate spatial plaque distribution patterns in coronary arteries based on computed tomography coronary angiography data sets and to express the learned patterns in prediction rules. An application is proposed to use these prediction rules for the detection of initially missed plaques.
Material and Methods: Two hundred fifty two consecutive patients with chronic coronary artery disease underwent contrast-enhanced dual-source computed tomography coronary angiography for clinical indications. Coronary artery plaques were manually labeled on a 16-segment coronary model and their position (ie, segments and bifurcations) and composition (ie, calcified, mixed, or noncalcified) were noted. The frequent itemset mining algorithm was used to statistically search for plaque distribution patterns. The patterns were expressed as prediction rules: given plaques at certain locations as conditions, a prediction rule gave evidence—with a certain confidence value—for a plaque at another location within the coronary artery tree. Prediction rules with the highest confidence values were evaluated and described. Furthermore, to improve manual plaque detection, all prediction rules were applied on the patient data to search for segments with potentially missed plaques. These segments were then reviewed in a second, guided reading for the existence of plaques. The same number of segments was also determined by a weighted random approach to evaluate the quality of prediction resulting from frequent itemset mining.
Results: In 200 of 252 (79.4%) patients, at least one coronary plaque (range, 1–22 plaques) was found. In total 1229 plaques (990 calcified, 80.6%; 227 mixed, 18.5%; 12 noncalcified, 1%) distributed, over 916 coronary segments and 507 vessels were manually labeled. Four plaque distribution patterns were identified: 20.6% of the patients had no plaques at all; 31.7% had plaques in the left coronary artery tree; 46.4% had plaques both in left and right coronary arteries, whereas 1.2% of the patients had plaques solely in the right coronary artery (RCA). General rules were found predicting plaques in the left anterior descending artery (LAD), given plaques in segments of the RCA or in the left main artery. Further general rules predicted plaques in the LAD, given plaques in the circumflex artery. In the guided review, the segment selection based on the prediction rules from frequent itemset mining performed significantly better (P < 0.001) than the weighted random approach by revealing 48 initially missed plaques.
Conclusions: This study demonstrates spatial plaque distribution patterns in coronary arteries as determined with cardiac CT. Use of the frequent itemset mining algorithm yielded rules that predicted plaques at certain sites given plaques at other sites of the coronary artery tree. Use of these prediction rules improved the manual labeling of coronary plaques as initially missed plaques could be predicted with the guided review.
DA - 2009///
PY - 2009
DO - http://dx.doi.org/10.1097/RLI.0b013e3181a8afc4
VL - 44
IS - 8
SP - 483
EP - 490
UR - http://journals.lww.com/investigativeradiology/Abstract/2009/08000/Prediction_Rules_for_the_Detection_of_Coronary.8.aspx
ER -

Objectives: To evaluate spatial plaque distribution patterns in coronary arteries based on computed tomography coronary angiography data sets and to express the learned patterns in prediction rules. An application is proposed to use these prediction rules for the detection of initially missed plaques. Material and Methods: Two hundred fifty two consecutive patients with chronic coronary artery disease underwent contrast-enhanced dual-source computed tomography coronary angiography for clinical indications. Coronary artery plaques were manually labeled on a 16-segment coronary model and their position (ie, segments and bifurcations) and composition (ie, calcified, mixed, or noncalcified) were noted. The frequent itemset mining algorithm was used to statistically search for plaque distribution patterns. The patterns were expressed as prediction rules: given plaques at certain locations as conditions, a prediction rule gave evidence—with a certain confidence value—for a plaque at another location within the coronary artery tree. Prediction rules with the highest confidence values were evaluated and described. Furthermore, to improve manual plaque detection, all prediction rules were applied on the patient data to search for segments with potentially missed plaques. These segments were then reviewed in a second, guided reading for the existence of plaques. The same number of segments was also determined by a weighted random approach to evaluate the quality of prediction resulting from frequent itemset mining. Results: In 200 of 252 (79.4%) patients, at least one coronary plaque (range, 1–22 plaques) was found. In total 1229 plaques (990 calcified, 80.6%; 227 mixed, 18.5%; 12 noncalcified, 1%) distributed, over 916 coronary segments and 507 vessels were manually labeled. Four plaque distribution patterns were identified: 20.6% of the patients had no plaques at all; 31.7% had plaques in the left coronary artery tree; 46.4% had plaques both in left and right coronary arteries, whereas 1.2% of the patients had plaques solely in the right coronary artery (RCA). General rules were found predicting plaques in the left anterior descending artery (LAD), given plaques in segments of the RCA or in the left main artery. Further general rules predicted plaques in the LAD, given plaques in the circumflex artery. In the guided review, the segment selection based on the prediction rules from frequent itemset mining performed significantly better (P < 0.001) than the weighted random approach by revealing 48 initially missed plaques. Conclusions: This study demonstrates spatial plaque distribution patterns in coronary arteries as determined with cardiac CT. Use of the frequent itemset mining algorithm yielded rules that predicted plaques at certain sites given plaques at other sites of the coronary artery tree. Use of these prediction rules improved the manual labeling of coronary plaques as initially missed plaques could be predicted with the guided review.

The Bayesian Group-Lasso for Analyzing Contingency Tables.

Sudhir Raman, Thomas J. Fuchs, Peter J. Wild, Edgar Dahl and Volker Roth.

Proceedings of the 26th Annual International Conference on Machine Learning, p. 881–888, ICML '09, ACM, New York, NY, USA, ISBN 978-1-60558-516-1, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{raman_bayesian_2009,
    address = {Montreal, Quebec, Canada},
    series = {{ICML} '09},
    title = {The {Bayesian} {Group}-{Lasso} for {Analyzing} {Contingency} {Tables}},
    isbn = {978-1-60558-516-1},
    url = {http://doi.acm.org/10.1145/1553374.1553487},
    doi = {10.1145/1553374.1553487},
    booktitle = {Proceedings of the 26th {Annual} {International} {Conference} on {Machine} {Learning}},
    publisher = {ACM, New York, NY, USA},
    author = {Raman, Sudhir and Fuchs, Thomas J. and Wild, Peter J. and Dahl, Edgar and Roth, Volker},
    year = {2009},
    pages = {881--888},
}

Download Endnote/RIS citation

TY - CONF
TI - The Bayesian Group-Lasso for Analyzing Contingency Tables
AU - Raman, Sudhir
AU - Fuchs, Thomas J.
AU - Wild, Peter J.
AU - Dahl, Edgar
AU - Roth, Volker
T3 - ICML '09
C1 - Montreal, Quebec, Canada
C3 - Proceedings of the 26th Annual International Conference on Machine Learning
DA - 2009///
PY - 2009
DO - 10.1145/1553374.1553487
SP - 881
EP - 888
PB - ACM, New York, NY, USA
SN - 978-1-60558-516-1
UR - http://doi.acm.org/10.1145/1553374.1553487
ER -

Inter-Active Learning of Randomized Tree Ensembles for Object Detection.

Thomas J. Fuchs and Joachim M. Buhmann.

Computer Vision Workshops (ICCV Workshops), 2009 IEEE 12th International Conference on Computer Vision, p. 1370–1377, ISBN 978-1-4244-4442-7, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuchs_inter-active_2009,
    title = {Inter-{Active} {Learning} of {Randomized} {Tree} {Ensembles} for {Object} {Detection}},
    isbn = {978-1-4244-4442-7},
    url = {http://dx.doi.org/10.1109/ICCVW.2009.5457452},
    doi = {http://dx.doi.org/10.1109/ICCVW.2009.5457452},
    booktitle = {Computer {Vision} {Workshops} ({ICCV} {Workshops}), 2009 {IEEE} 12th {International} {Conference} on {Computer} {Vision}},
    author = {Fuchs, Thomas J. and Buhmann, Joachim M.},
    year = {2009},
    pages = {1370--1377},
}

Download Endnote/RIS citation

TY - CONF
TI - Inter-Active Learning of Randomized Tree Ensembles for Object Detection
AU - Fuchs, Thomas J.
AU - Buhmann, Joachim M.
C3 - Computer Vision Workshops (ICCV Workshops), 2009 IEEE 12th International Conference on Computer Vision
DA - 2009///
PY - 2009
DO - http://dx.doi.org/10.1109/ICCVW.2009.5457452
SP - 1370
EP - 1377
SN - 978-1-4244-4442-7
UR - http://dx.doi.org/10.1109/ICCVW.2009.5457452
ER -

Guided Review by Frequent Itemset Mining: Additional Evidence for Plaque Detection.

Stefan C. Saur, Hatem Alkadhi, Lotus Desbiolles, Thomas J. Fuchs, Gabor Szekely and Philippe C. Cattin.

International Journal of Computer Assisted Radiology and Surgery, vol. 4, 3, p. 263–271, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{saur_guided_2009,
    title = {Guided {Review} by {Frequent} {Itemset} {Mining}: {Additional} {Evidence} for {Plaque} {Detection}},
    volume = {4},
    url = {http://dx.doi.org/10.1007/s11548-009-0290-5},
    number = {3},
    journal = {International Journal of Computer Assisted Radiology and Surgery},
    author = {Saur, Stefan C. and Alkadhi, Hatem and Desbiolles, Lotus and Fuchs, Thomas J. and Szekely, Gabor and Cattin, Philippe C.},
    year = {2009},
    pages = {263--271},
}

Download Endnote/RIS citation

TY - JOUR
TI - Guided Review by Frequent Itemset Mining: Additional Evidence for Plaque Detection
AU - Saur, Stefan C.
AU - Alkadhi, Hatem
AU - Desbiolles, Lotus
AU - Fuchs, Thomas J.
AU - Szekely, Gabor
AU - Cattin, Philippe C.
T2 - International Journal of Computer Assisted Radiology and Surgery
DA - 2009///
PY - 2009
VL - 4
IS - 3
SP - 263
EP - 271
UR - http://dx.doi.org/10.1007/s11548-009-0290-5
ER -

Randomized Tree Ensembles for Object Detection in Computational Pathology.

Thomas J. Fuchs, Johannes Haybaeck, Peter J. Wild, Mathias Heikenwalder, Holger Moch, Adriano Aguzzi and Joachim M. Buhmann.

Proceedings of the 5th International Symposium on Advances in Visual Computing: Part I, p. 367–378, ISVC '09, Springer-Verlag, Berlin, Heidelberg, ISBN 978-3-642-10330-8, 2009

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuchs_randomized_2009,
    address = {Las Vegas, Nevada},
    series = {{ISVC} '09},
    title = {Randomized {Tree} {Ensembles} for {Object} {Detection} in {Computational} {Pathology}},
    isbn = {978-3-642-10330-8},
    url = {http://dx.doi.org/10.1007/978-3-642-10331-5_35},
    doi = {http://dx.doi.org/10.1007/978-3-642-10331-5_35},
    booktitle = {Proceedings of the 5th {International} {Symposium} on {Advances} in {Visual} {Computing}: {Part} {I}},
    publisher = {Springer-Verlag, Berlin, Heidelberg},
    author = {Fuchs, Thomas J. and Haybaeck, Johannes and Wild, Peter J. and Heikenwalder, Mathias and Moch, Holger and Aguzzi, Adriano and Buhmann, Joachim M.},
    year = {2009},
    pages = {367--378},
}

Download Endnote/RIS citation

TY - CONF
TI - Randomized Tree Ensembles for Object Detection in Computational Pathology
AU - Fuchs, Thomas J.
AU - Haybaeck, Johannes
AU - Wild, Peter J.
AU - Heikenwalder, Mathias
AU - Moch, Holger
AU - Aguzzi, Adriano
AU - Buhmann, Joachim M.
T3 - ISVC '09
C1 - Las Vegas, Nevada
C3 - Proceedings of the 5th International Symposium on Advances in Visual Computing: Part I
DA - 2009///
PY - 2009
DO - http://dx.doi.org/10.1007/978-3-642-10331-5_35
SP - 367
EP - 378
PB - Springer-Verlag, Berlin, Heidelberg
SN - 978-3-642-10330-8
UR - http://dx.doi.org/10.1007/978-3-642-10331-5_35
ER -

Automatic and Robust Forearm Segmentation Using Graph Cuts.

Philipp Fuernstahl, Thomas J. Fuchs, Andreas Schweizer, Ladislav Nagy, Gabor Szekely and Matthias Harders.

5th IEEE International Symposium on Biomedical Imaging. ISBI 2008, p. 77–80, IEEE, ISBN 978-1-4244-2002-5, 2008

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuernstahl_automatic_2008,
    address = {Paris, France},
    title = {Automatic and {Robust} {Forearm} {Segmentation} {Using} {Graph} {Cuts}},
    isbn = {978-1-4244-2002-5},
    url = {http://dx.doi.org/10.1109/ISBI.2008.4540936},
    doi = {10.1109/ISBI.2008.4540936},
    abstract = {The segmentation of bones in computed tomography (CT) images is an important step for the simulation of forearm bone motion, since it allows to include patient specific anatomy in a kinematic model. While the identification of the bone diaphysis is straightforward, the segmentation of bone joints with weak, thin, and diffusive boundaries is still a challenge. We propose a graph cut segmentation approach that is particularly suited to robustly segment joints in 3-d CT images. We incorporate knowledge about intensity, bone shape and local structures into a novel energy function. Our presented framework performs a simultaneous segmentation of both forearm bones without any user interaction.},
    booktitle = {5th {IEEE} {International} {Symposium} on {Biomedical} {Imaging}. {ISBI} 2008},
    publisher = {IEEE},
    author = {Fuernstahl, Philipp and Fuchs, Thomas J. and Schweizer, Andreas and Nagy, Ladislav and Szekely, Gabor and Harders, Matthias},
    year = {2008},
    pages = {77--80},
}

Download Endnote/RIS citation

TY - CONF
TI - Automatic and Robust Forearm Segmentation Using Graph Cuts
AU - Fuernstahl, Philipp
AU - Fuchs, Thomas J.
AU - Schweizer, Andreas
AU - Nagy, Ladislav
AU - Szekely, Gabor
AU - Harders, Matthias
AB - The segmentation of bones in computed tomography (CT) images is an important step for the simulation of forearm bone motion, since it allows to include patient specific anatomy in a kinematic model. While the identification of the bone diaphysis is straightforward, the segmentation of bone joints with weak, thin, and diffusive boundaries is still a challenge. We propose a graph cut segmentation approach that is particularly suited to robustly segment joints in 3-d CT images. We incorporate knowledge about intensity, bone shape and local structures into a novel energy function. Our presented framework performs a simultaneous segmentation of both forearm bones without any user interaction.
C1 - Paris, France
C3 - 5th IEEE International Symposium on Biomedical Imaging. ISBI 2008
DA - 2008///
PY - 2008
DO - 10.1109/ISBI.2008.4540936
SP - 77
EP - 80
PB - IEEE
SN - 978-1-4244-2002-5
UR - http://dx.doi.org/10.1109/ISBI.2008.4540936
ER -

The segmentation of bones in computed tomography (CT) images is an important step for the simulation of forearm bone motion, since it allows to include patient specific anatomy in a kinematic model. While the identification of the bone diaphysis is straightforward, the segmentation of bone joints with weak, thin, and diffusive boundaries is still a challenge. We propose a graph cut segmentation approach that is particularly suited to robustly segment joints in 3-d CT images. We incorporate knowledge about intensity, bone shape and local structures into a novel energy function. Our presented framework performs a simultaneous segmentation of both forearm bones without any user interaction.

Weakly Supervised Cell Nuclei Detection and Segmentation on Tissue Microarrays of Renal Cell Carcinoma.

Thomas J. Fuchs, Tilman Lange, Peter J. Wild, Holger Moch and Joachim M. Buhmann.

Pattern Recognition. DAGM 2008, vol. 5096, p. 173–182, Lecture Notes in Computer Science, Springer-Verlag, ISBN 978-3-540-69320-8, 2008

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuchs_weakly_2008,
    address = {Berlin, Heidelberg},
    series = {Lecture {Notes} in {Computer} {Science}},
    title = {Weakly {Supervised} {Cell} {Nuclei} {Detection} and {Segmentation} on {Tissue} {Microarrays} of {Renal} {Cell} {Carcinoma}},
    volume = {5096},
    isbn = {978-3-540-69320-8},
    url = {http://www.springerlink.com/index/L380G56T1V230715.pdf},
    doi = {10.1007/978-3-540-69321-},
    booktitle = {Pattern {Recognition}. {DAGM} 2008},
    publisher = {Springer-Verlag},
    author = {Fuchs, Thomas J. and Lange, Tilman and Wild, Peter J. and Moch, Holger and Buhmann, Joachim M.},
    year = {2008},
    pages = {173--182},
}

Download Endnote/RIS citation

TY - CONF
TI - Weakly Supervised Cell Nuclei Detection and Segmentation on Tissue Microarrays of Renal Cell Carcinoma
AU - Fuchs, Thomas J.
AU - Lange, Tilman
AU - Wild, Peter J.
AU - Moch, Holger
AU - Buhmann, Joachim M.
T3 - Lecture Notes in Computer Science
C1 - Berlin, Heidelberg
C3 - Pattern Recognition. DAGM 2008
DA - 2008///
PY - 2008
DO - 10.1007/978-3-540-69321-
VL - 5096
SP - 173
EP - 182
PB - Springer-Verlag
SN - 978-3-540-69320-8
UR - http://www.springerlink.com/index/L380G56T1V230715.pdf
ER -

Computational Pathology Analysis of Tissue Microarrays Predicts Survival of Renal Clear Cell Carcinoma Patients.

Thomas J. Fuchs, Peter J. Wild, Holger Moch and Joachim M. Buhmann.

Proceedings of the international conference on Medical Image Computing and Computer-Assisted Intervention MICCAI, vol. 5242, p. 1-8, Lecture Notes in Computer Science, Springer-Verlag, ISBN 978-3-540-85989-5, 2008

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@inproceedings{fuchs_computational_2008,
    address = {Berlin, Heidelberg},
    series = {Lecture {Notes} in {Computer} {Science}},
    title = {Computational {Pathology} {Analysis} of {Tissue} {Microarrays} {Predicts} {Survival} of {Renal} {Clear} {Cell} {Carcinoma} {Patients}},
    volume = {5242},
    isbn = {978-3-540-85989-5},
    url = {http://dx.doi.org/10.1007/978-3-540-85990-1_1},
    doi = {10.1007/978-3-540-85990-1_1},
    booktitle = {Proceedings of the international conference on {Medical} {Image} {Computing} and {Computer}-{Assisted} {Intervention} {MICCAI}},
    publisher = {Springer-Verlag},
    author = {Fuchs, Thomas J. and Wild, Peter J. and Moch, Holger and Buhmann, Joachim M.},
    year = {2008},
    keywords = {Computer Science},
    pages = {1--8},
}

Download Endnote/RIS citation

TY - CONF
TI - Computational Pathology Analysis of Tissue Microarrays Predicts Survival of Renal Clear Cell Carcinoma Patients
AU - Fuchs, Thomas J.
AU - Wild, Peter J.
AU - Moch, Holger
AU - Buhmann, Joachim M.
T3 - Lecture Notes in Computer Science
C1 - Berlin, Heidelberg
C3 - Proceedings of the international conference on Medical Image Computing and Computer-Assisted Intervention MICCAI
DA - 2008///
PY - 2008
DO - 10.1007/978-3-540-85990-1_1
VL - 5242
SP - 1
EP - 8
PB - Springer-Verlag
SN - 978-3-540-85989-5
UR - http://dx.doi.org/10.1007/978-3-540-85990-1_1
KW - Computer Science
ER -

Nuclear Karyopherin Alpha2 Expression Predicts Poor Survival in Patients with Advanced Breast Cancer Irrespective of Treatment Intensity.

Oleg Gluz, Peter Wild, Robert Meiler, Raihana Diallo-Danebrock, Evelyn Ting, Svjetlana Mohrmann, Gerhart Schuett, Edgar Dahl, Thomas J. Fuchs, Alexander Herr, Andreas Gaumann, Markus Frick, Christopher Poremba, Ulrike Anneliese Nitz and Arndt Hartmann.

International Journal of Cancer, vol. 123/6, IF 4,693, p. 1433–1438, 2008

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{gluz_nuclear_2008,
    title = {Nuclear {Karyopherin} {Alpha2} {Expression} {Predicts} {Poor} {Survival} in {Patients} with {Advanced} {Breast} {Cancer} {Irrespective} of {Treatment} {Intensity}},
    volume = {123/6},
    issn = {1097-0215},
    url = {http://onlinelibrary.wiley.com/doi/10.1002/ijc.23628/full},
    doi = {10.1002/ijc.23628},
    number = {IF 4,693},
    journal = {International Journal of Cancer},
    author = {Gluz, Oleg and Wild, Peter and Meiler, Robert and Diallo-Danebrock, Raihana and Ting, Evelyn and Mohrmann, Svjetlana and Schuett, Gerhart and Dahl, Edgar and Fuchs, Thomas J. and Herr, Alexander and Gaumann, Andreas and Frick, Markus and Poremba, Christopher and Nitz, Ulrike Anneliese and Hartmann, Arndt},
    year = {2008},
    pages = {1433--1438},
}

Download Endnote/RIS citation

TY - JOUR
TI - Nuclear Karyopherin Alpha2 Expression Predicts Poor Survival in Patients with Advanced Breast Cancer Irrespective of Treatment Intensity
AU - Gluz, Oleg
AU - Wild, Peter
AU - Meiler, Robert
AU - Diallo-Danebrock, Raihana
AU - Ting, Evelyn
AU - Mohrmann, Svjetlana
AU - Schuett, Gerhart
AU - Dahl, Edgar
AU - Fuchs, Thomas J.
AU - Herr, Alexander
AU - Gaumann, Andreas
AU - Frick, Markus
AU - Poremba, Christopher
AU - Nitz, Ulrike Anneliese
AU - Hartmann, Arndt
T2 - International Journal of Cancer
DA - 2008///
PY - 2008
DO - 10.1002/ijc.23628
VL - 123/6
IS - IF 4,693
SP - 1433
EP - 1438
SN - 1097-0215
UR - http://onlinelibrary.wiley.com/doi/10.1002/ijc.23628/full
ER -

The Acute Coronary Syndrome – Pre-Hospital Diagnostic Quality.

Geza Gemes, Thomas J. Fuchs, Gernot Wildner, Freyja-Maria Smolle-Juettner, Josef Smolle, Kurt Stoschitzky and Gerhard Prause.

Resuscitation, vol. 66, 3, p. 323 - 330, 2005

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

@article{gemes_acute_2005,
    title = {The {Acute} {Coronary} {Syndrome} – {Pre}-{Hospital} {Diagnostic} {Quality}},
    volume = {66},
    issn = {0300-9572},
    url = {http://www.sciencedirect.com/science/article/B6T19-4GV9846-3/2/bcd3755266ba68b84536b39803426401},
    doi = {DOI: 10.1016/j.resuscitation.2005.04.006},
    abstract = {Background and objective: In the Austrian emergency medical service (EMS), emergency medical technician-staffed and physician-staffed vehicles are in operation. Patients with suspected acute coronary syndromes (ACS) are treated in the pre-hospital phase and transported to the hospital by an emergency physician (EP). This study evaluates the diagnostic performance of EPs in ACS and the impact of this emergency system on the outcome of ACS in an urban area.
    Design: Retrospective case control study.
    Methods: All protocol sheets from the emergency physicians were searched for the diagnosis of ACS. The database of the emergency department (ED) was searched for patients with ACS as an admission diagnosis or ACS as discharge diagnosis. For patients admitted to an intensive care unit (ICU), the medical history from the ICU was reviewed. According to the diagnosis and the aggressiveness of therapy, patients were divided in five categories of severity at each stage of care (pre-hospital category, ED category, ICU category).
    Results: A total of 3585 patients was analysed. Only 17.8\% of the patients with ACS as the admission diagnosis and 20.3\% of the patients with ACS as the discharge diagnosis were transported by an EP. 46.8\% of the ACS diagnosis by EPs were confirmed in hospital. Patients transported by EPs showed a higher all-cause mortality in hospital (1.6\% vs. 0.6\%; p = 0.011). There was no significant correlation between the pre-hospital category of patients treated by EPs and the ED category. When a 12-lead-electrocardiogramm was recorded, the correlation improved slightly (rho: 0.139; p = 0.006).
    Conclusions: The percentage of ACS patients transported to hospital by an EP is very low, and EPs seem to be “over-aware” in the diagnosis of ACS.},
    number = {3},
    journal = {Resuscitation},
    author = {Gemes, Geza and Fuchs, Thomas J. and Wildner, Gernot and Smolle-Juettner, Freyja-Maria and Smolle, Josef and Stoschitzky, Kurt and Prause, Gerhard},
    year = {2005},
    pages = {323 -- 330},
}

Download Endnote/RIS citation

TY - JOUR
TI - The Acute Coronary Syndrome – Pre-Hospital Diagnostic Quality
AU - Gemes, Geza
AU - Fuchs, Thomas J.
AU - Wildner, Gernot
AU - Smolle-Juettner, Freyja-Maria
AU - Smolle, Josef
AU - Stoschitzky, Kurt
AU - Prause, Gerhard
T2 - Resuscitation
AB - Background and objective: In the Austrian emergency medical service (EMS), emergency medical technician-staffed and physician-staffed vehicles are in operation. Patients with suspected acute coronary syndromes (ACS) are treated in the pre-hospital phase and transported to the hospital by an emergency physician (EP). This study evaluates the diagnostic performance of EPs in ACS and the impact of this emergency system on the outcome of ACS in an urban area.
Design: Retrospective case control study.
Methods: All protocol sheets from the emergency physicians were searched for the diagnosis of ACS. The database of the emergency department (ED) was searched for patients with ACS as an admission diagnosis or ACS as discharge diagnosis. For patients admitted to an intensive care unit (ICU), the medical history from the ICU was reviewed. According to the diagnosis and the aggressiveness of therapy, patients were divided in five categories of severity at each stage of care (pre-hospital category, ED category, ICU category).
Results: A total of 3585 patients was analysed. Only 17.8% of the patients with ACS as the admission diagnosis and 20.3% of the patients with ACS as the discharge diagnosis were transported by an EP. 46.8% of the ACS diagnosis by EPs were confirmed in hospital. Patients transported by EPs showed a higher all-cause mortality in hospital (1.6% vs. 0.6%; p = 0.011). There was no significant correlation between the pre-hospital category of patients treated by EPs and the ED category. When a 12-lead-electrocardiogramm was recorded, the correlation improved slightly (rho: 0.139; p = 0.006).
Conclusions: The percentage of ACS patients transported to hospital by an EP is very low, and EPs seem to be “over-aware” in the diagnosis of ACS.
DA - 2005///
PY - 2005
DO - DOI: 10.1016/j.resuscitation.2005.04.006
VL - 66
IS - 3
SP - 323
EP - 330
SN - 0300-9572
UR - http://www.sciencedirect.com/science/article/B6T19-4GV9846-3/2/bcd3755266ba68b84536b39803426401
ER -

Background and objective: In the Austrian emergency medical service (EMS), emergency medical technician-staffed and physician-staffed vehicles are in operation. Patients with suspected acute coronary syndromes (ACS) are treated in the pre-hospital phase and transported to the hospital by an emergency physician (EP). This study evaluates the diagnostic performance of EPs in ACS and the impact of this emergency system on the outcome of ACS in an urban area. Design: Retrospective case control study. Methods: All protocol sheets from the emergency physicians were searched for the diagnosis of ACS. The database of the emergency department (ED) was searched for patients with ACS as an admission diagnosis or ACS as discharge diagnosis. For patients admitted to an intensive care unit (ICU), the medical history from the ICU was reviewed. According to the diagnosis and the aggressiveness of therapy, patients were divided in five categories of severity at each stage of care (pre-hospital category, ED category, ICU category). Results: A total of 3585 patients was analysed. Only 17.8% of the patients with ACS as the admission diagnosis and 20.3% of the patients with ACS as the discharge diagnosis were transported by an EP. 46.8% of the ACS diagnosis by EPs were confirmed in hospital. Patients transported by EPs showed a higher all-cause mortality in hospital (1.6% vs. 0.6%; p = 0.011). There was no significant correlation between the pre-hospital category of patients treated by EPs and the ED category. When a 12-lead-electrocardiogramm was recorded, the correlation improved slightly (rho: 0.139; p = 0.006). Conclusions: The percentage of ACS patients transported to hospital by an EP is very low, and EPs seem to be “over-aware” in the diagnosis of ACS.

Efficient Visualization of Whole Slide Images in Web-based Viewers for Digital Pathology.

Peter J. Schüffler, Evangelos Stamelos, Ishtiaque Ahmed, D. Vijay K. Yarlagadda, Matthew G. Hanna, Victor E. Reuter, David S. Klimstra and Meera Hameed.

Archives of Pathology & Laboratory Medicine, 2022-01-3

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Integrated digital pathology at scale: A solution for clinical diagnostics and cancer research at a large academic medical center.

Journal of the American Medical Informatics Association, vol. 28, 9, p. 1874-1884, July 14, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Digital Pathology Operations at an NYC Tertiary Cancer Center During the First 4 Months of COVID-19 Pandemic Response.

Academic Pathology, vol. 8, p. 23742895211010276, April 28, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Multi-magnification-based machine learning as an ancillary tool for the pathologic assessment of shaved margins for breast carcinoma lumpectomy specimens.

Timothy M. D’Alfonso, David Joon Ho, Matthew G. Hanna, Anne Grabenstetter, Dig Vijay Kumar Yarlagadda, Luke Geneslaw, Peter Ntiamoah, Thomas J. Fuchs and Lee K. Tan.

Modern Pathology, 4-26-2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Overcoming an annotation hurdle: Digitizing pen annotations from whole slide images.

Peter J. Schüffler, Dig Vijay Kumar Yarlagadda, Chad Vanderbilt and Thomas J. Fuchs.

Journal of Pathology Informatics, vol. 12, 1, p. 9, 02/23/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Deep Multi-Magnification Networks for Multi-Class Breast Cancer Image Segmentation.

David Joon Ho, Dig V. K. Yarlagadda, Timothy M. D'Alfonso, Matthew G. Hanna, Anne Grabenstetter, Peter Ntiamoah, Edi Brogi, Lee K. Tan and Thomas J. Fuchs.

Computerized Medical Imaging and Graphics, vol. 88, 1/12/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Deep Learning Methods for Lung Cancer Segmentation in Whole-Slide Histopathology Images—The ACDC@LungHP Challenge 2019.

IEEE J. Biomed. Health Inform., vol. 25, 2, p. 429-440, 2/2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Flextilesource: An openseadragon extension for efficient whole-slide image visualization.

Peter J. Schüffler, Gamze Gokturk Ozcan, Hikmat Al-Ahmadie and Thomas J. Fuchs.

J Pathol Inform, vol. 12, 1, p. 31, 2021

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Beyond Classification: Whole Slide Tissue Histopathology Analysis By End-To-End Part Learning.

Chensu Xie, Hassan Muhammad, Chad M. Vanderbilt, Raul Caso, Dig Vijay Kumar Yarlagadda, Gabriele Campanella and Thomas J. Fuchs.

Proceedings of the Third Conference on Medical Imaging with Deep Learning, Medical Imaging with Deep Learning, vol. 121, p. 843-856, PMLR, 6-24-2020

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Validation of a digital pathology system including remote review during the COVID-19 pandemic.

Modern Pathology, vol. 33, p. 2115–2127, 2020-06-22

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

(Re) Defining the high-power field for digital pathology.

David Kim, Liron Pantanowitz, Peter Schüffler, Dig Vijay Kumar Yarlagadda, Orly Ardon, Victor E. Reuter, Meera Hameed, David S. Klimstra and Matthew G. Hanna.

Journal of Pathology Informatics, vol. 11, 1, p. 33, 1/1/2020

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Deep Interactive Learning: An Efficient Labeling Approach for Deep Learning-Based Osteosarcoma Treatment Response Assessment.

David Joon Ho, Narasimhan P. Agaram, Peter J. Schüffler, Chad M. Vanderbilt, Marc-Henri Jean, Meera R. Hameed and Thomas J. Fuchs.

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Unsupervised Subtyping of Cholangiocarcinoma Using a Deep Clustering Convolutional Autoencoder.

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Accuracy of Intraoperative Frozen Section of Sentinel Lymph Nodes After Neoadjuvant Chemotherapy for Breast Carcinoma.

Am. J. Surg. Pathol., Jun 18, 2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

VOCA: Cell Nuclei Detection In Histopathology Images By Vector Oriented Confidence Accumulation.

Chensu Xie, Chad M. Vanderbilt, Anne Grabenstetter and Thomas J. Fuchs.

Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, MIDL, vol. 102, p. 527-539, PMLR, 5/23/2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

DeepPET: A deep encoder–decoder network for directly solving the PET image reconstruction inverse problem.

Ida Häggström, C. Ross Schmidtlein, Gabriele Campanella and Thomas J. Fuchs.

Medical Image Analysis, vol. 54, p. 253-262, 2019-03-30

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Towards Unsupervised Cancer Subtyping: Predicting Prognosis Using A Histologic Visual Dictionary.

arXiv:1903.05257 [cs, q-bio, stat], 2019-03-12

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Whole slide imaging equivalency and efficiency study: experience at a large academic center.

Modern Pathology, vol. 32, p. 916–928, 2019-02-18

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Clinical-grade computational pathology using weakly supervised deep learning on whole slide images.

Gabriele Campanella, Matthew G. Hanna, Luke Geneslaw, Allen Miraflor, Vitor Werneck Krauss Silva, Klaus J. Busam, Edi Brogi, Victor E. Reuter, David S. Klimstra and Thomas J. Fuchs.

Nat Med, vol. 25, 8, p. 1301-1309, 8/2019

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Comparison of contrast-enhanced and diffusion-weighted MRI in assessment of the terminal ileum in Crohn’s disease patients.

Carl A. J. Puylaert, Jeroen A. W. Tielbeek, Peter J. Schüffler, C. Yung Nio, Karin Horsthuis, Banafsche Mearadji, Cyriel Y. Ponsioen, Frans M. Vos and Jaap Stoker.

Abdominal Radiology, vol. 44, p. 398–405, 2018-8-14

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Terabyte-scale Deep Multiple Instance Learning for Classification and Localization in Pathology.

Gabriele Campanella, Vitor Werneck Krauss Silva and Thomas J. Fuchs.

arXiv:1805.06983 [cs], 2018-05-17

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

DeepPET: A deep encoder–decoder network for directly solving the PET reconstruction inverse problem.

Ida Häggström, C. Ross Schmidtlein, Gabriele Campanella and Thomas J. Fuchs.

arXiv:1804.0785 [cs.CV], 2018-04-20

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Comprehensive immunohistochemical analysis of PD-L1 shows scarce expression in castration-resistant prostate cancer.

Christian D. Fankhauser, Peter J. Schüffler, Silke Gillessen, Aurelius Omlin, Niels J. Rupp, Jan H. Rueschoff, Thomas Hermanns, Cedric Poyet, Tullio Sulser, Holger Moch and Peter J. Wild.

Oncotarget, vol. 9, 12, 2018-02-13

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Towards machine learned quality control: A benchmark for sharpness quantification in digital pathology.

Gabriele Campanella, Arjun R. Rajanna, Lorraine Corsale, Peter J. Schüffler, Yukako Yagi and Thomas J. Fuchs.

Computerized Medical Imaging and Graphics, vol. 65, p. 142-151, 04/2018

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Semiautomatic Assessment of the Terminal Ileum and Colon in Patients with Crohn Disease Using MRI (the VIGOR++ Project).

Academic Radiology, vol. 25, 8, p. 1038-1045, 2/2018

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Computational Pathology.

Peter J. Schüffler, Qing Zhong, Peter J. Wild and Thomas J. Fuchs.

In: Johannes Haybäck (ed.) Mechanisms of Molecular Carcinogenesis - Volume 2, 1st ed. 2017 edition, Springer, ISBN 3-319-53660-5, 21. Jun. 2017

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Scaling of cytoskeletal organization with cell size in Drosophila.

Alison K. Spencer, Andrew J. Schaumberg and Jennifer A. Zallen.

Mol. Biol. Cell, p. mbc.E16-10-0691, 2017-04-12

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Hybrid Deep Learning on Single Wide-field Optical Coherence Tomography Scans Accurately Classifies Glaucoma Suspects.

Hassan Muhammad, Thomas J. Fuchs, Nicole De Cuir, Carlos G. De Moraes, Dana M. Blumberg, Jeffrey M. Liebmann, Robert Ritch and Donald C. Hood.

Journal of Glaucoma, vol. 26, 12, 10/2017

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

MRI-Based Surgical Planning for Lumbar Spinal Stenosis.

Gabriele Abbati, Stefan Bauer, Sebastian Winklhofer, Peter J. Schüffler, Ulrike Held, Jakob M. Burgstaller, Johann Steurer and Joachim M. Buhmann.

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations.

Peter J. Schüffler, Judy Sarungbam, Hassan Muhammad, Ed Reznik, Satish K. Tickoo and Thomas J. Fuchs.

Proceedings of the 1st Machine Learning for Healthcare Conference, Machine Learning for Healthcare, vol. 56, p. 191-208, Proceedings of Machine Learning Research, PMLR, 18 Aug 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

A generalized computer vision approach to mapping crop fields in heterogeneous agricultural landscapes.

Stephanie R. Debats, Dee Luo, Lyndon D. Estes, Thomas J. Fuchs and Kelly K. Caylor.

Remote Sensing of Environment, vol. 179, p. 210-221, June 15, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Is There an Association Between Pain and Magnetic Resonance Imaging Parameters in Patients With Lumbar Spinal Stenosis?

Spine, vol. 41, 17, p. 1053-1062, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

DeepScope: Nonintrusive Whole Slide Saliency Annotation and Prediction from Pathologists at the Microscope.

Andrew J. Schaumberg, S. Joseph Sirintrapun, Hikmat A. Al-Ahmadie, Peter J. Schüffler and Thomas J. Fuchs.

13th International Conference on Computational Intelligence methods for Bioinformatics and Biostatistics, CIBB, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Big Data in der Medizin.

Qing Zhong, Roman Barnert, Gunnar Ratsch, Thomas J. Fuchs and Peter J. Wild.

Leading Opinions: Hämatologie & Onkologie, p. 102-105, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Cancer-secreted AGR2 induces programmed cell death in normal cells.

Elizabeth A. Vitello, Sue-Ing Quek, Heather Kincaid, Thomas Fuchs, Daniel J. Crichton, Pamela Troisch and Alvin Y. Liu.

Oncotarget, vol. 5, 0, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Multi-Organ Cancer Classification and Survival Analysis.

Stefan Bauer, Nicolas Carion, Peter Schüffler, Thomas Fuchs, Peter Wild and Joachim M. Buhmann.

arXiv:1606.00897 [cs, q-bio, stat], 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Classifying Renal Cell Carcinoma by Using Convolutional Neural Networks to Deconstruct Pathological Images.

Hassan Muhammad, Peter J. Schüffler, Judy Sarungbam, Satish K. Tickoo and Thomas Fuchs.

Vincent du Vigneaud Memorial Research Symposium, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Image-Based Computational Quantification and Visualization of Genetic Alterations and Tumour Heterogeneity.

Scientific Reports, vol. 6, p. 24146, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Autonomous Terrain Classification With Co- and Self-Training Approach.

K. Otsu, M. Ono, T. J. Fuchs, I. Baldwin and T. Kubota.

IEEE Robotics and Automation Letters, vol. 1, 2, p. 814-819, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Predicting Drug Interactions and Mutagenicity with Ensemble Classifiers on Subgraphs of Molecules.

Andrew Schaumberg, Angela Yu, Tatsuhiro Koshi, Xiaochan Zong and Santoshkalyan Rayadhurgam.

arXiv:1601.07233 [cs, stat], 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Real-Time Data Mining of Massive Data Streams from Synoptic Sky Surveys.

S. G. Djorgovski, M. J. Graham, C. Donalek, A. A. Mahabal, A. J. Drake, M. Turmon and T.J. Fuchs.

Future Generation Computer Systems, 2016

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Automatic single cell segmentation on highly multiplexed tissue images.

Peter J. Schüffler, Denis Schapiro, Charlotte Giesen, Hao A. O. Wang, Bernd Bodenmiller and Joachim M. Buhmann.

Cytometry Part A, vol. 87, 10, p. 936-942, 10/2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Understanding Neural Networks Through Deep Visualization.

Jason Yosinski, Jeff Clune, Anh Nguyen, Thomas Fuchs and Hod Lipson.

ICML Deep Learning Workshop, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Enhanced Flyby Science with Onboard Computer Vision: Tracking and Surface Feature Detection at Small Bodies.

Thomas J. Fuchs, David R. Thompson, Brian D. Bue, Julie Castillo-Rogez, Steve A. Chien, Dero Gharibian and Kiri L. Wagstaff.

Earth and Space Science, vol. 2, 10, p. 417-34, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Boosting Convolutional Features for Robust Object Proposals.

Nikolaos Karianakis, Thomas J. Fuchs and Stefano Soatto.

2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Machine Learning Approaches to Analyze Histological Images of Tissues from Radical Prostatectomies.

Arkadiusz Gertych, Nathan Ing, Zhaoxuan Ma, Thomas J. Fuchs, Sadri Salman, Sambit Mohanty, Sanica Bhele, Adriana Velásquez-Vacca, Mahul B. Amin and Beatrice S. Knudsen.

Computerized Medical Imaging and Graphics, vol. 46, Part 2, p. 197-208, Information Technologies in Biomedicine, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Risk-aware Planetary Rover Operation: Autonomous Terrain Classification and Path Planning.

Masahiro Ono, Thomas J. Fuchs, Amanda Steffy, Mark Maimone and Jeng Yen.

Proceedings of the 36th IEEE Aerospace Conference, p. 1–10, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Robot-Centric Activity Prediction from First-Person Videos: What Will They Do to Me?

Michael S. Ryoo, Thomas J. Fuchs, Lu Xia, J. K. Aggarwal and Larry H. Matthies.

Proceedings of the 10th ACM/IEEE International Conference on Human-Robot Interaction, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Crohn's Disease Segmentation from MRI Using Learned Image Priors.

D. Mahapatra, P. J. Schüffler, F. M. Vos and J. M. Buhmann.

Proceedings IEEE ISBI 2015, p. 625-628, 2015

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Autonomous Real-time Detection of Plumes and Jets from Moons and Comets.

Kiri L. Wagstaff, David R. Thompson, Brian D. Bue and Thomas J. Fuchs.

ApJ, vol. 794, 1, p. 43, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Sparse Meta-Gaussian Information Bottleneck.

Melanie Rey, Thomas J. Fuchs and Volker Roth.

Proceedings of the 31st International Conference on Machine Learning, ICML, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

KPNA2 Is Overexpressed in Human and Mouse Endometrial Cancers and Promotes Cellular Proliferation.

The Journal of Pathology, vol. 234, 2, p. 239–252, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Highly Multiplexed Imaging of Tumor Tissues with Subcellular Resolution by Mass Cytometry.

C. Giesen, H. A. Wang, D. Schapiro, N. Zivanovic, A. Jacobs, B. Hattendorf, P. J. Schüffler, D. Grolimund, J. M. Buhmann, S. Brandt, Z. Varga, P. J. Wild, D. Gunther and B. Bodenmiller.

Nature methods, vol. 11, p. 417-22, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Automated Real-Time Classification and Decision Making in Massive Data Streams from Synoptic Sky Surveys.

S.G. Djorgovski, A. Mahabal, C. Donalek, M. Graham, A. Drake, M. Turmon and T.J. Fuchs.

2014 IEEE 10th International Conference on e-Science (e-Science), 2014 IEEE 10th International Conference on e-Science (e-Science), vol. 1, p. 204-211, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Early Recognition of Human Activities from First-Person Videos Using Onset Representations.

M. S. Ryoo, Thomas J. Fuchs, Lu Xia, J. K. Aggarwal and Larry Matthies.

arXiv:1406.5309 [cs], 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Computer Aided Crohn's Disease Severity Assessment in MRI.

Peter J. Schüffler, Dwarikanath Mahapatra, Franciscus M. Vos and Joachim M. Buhmann.

VIGOR++ Workshop 2014 - Showcase of Research Outcomes and Future Outlook, 2014

Best Poster Award

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Semi-Automatic Crohn's Disease Severity Estimation on MR Imaging.

P. J. Schüffler, D. Mahapatra, R. E. Naziroglu, Z. Li, C. A. J. Puylaert, R. Andriantsimiavona, F. M. Vos, D. A. Pendsé, C. Yung Nio, J. Stoker, S. A. Taylor and J. M. Buhmann.

6th MICCAI Workshop on Abdominal Imaging - Computational and Clinical Applications, 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Autonomous Onboard Surface Feature Detection for Flyby Missions.

Thomas J. Fuchs, Brian D. Bue, Julie Castillo-Rogez, Steve A. Chien, Kiri Wagstaff and David R. Thompson.

Proceedings of the 12th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS), 2014

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

TMARKER: A Free Software Toolkit for Histopathological Cell Counting and Staining Estimation.

Peter J. Schüffler, Thomas J. Fuchs, Cheng S. Ong, Peter Wild and Joachim M. Buhmann.

Journal of Pathology Informatics, vol. 4, 2, p. 2, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Smart, Texture-Sensitive Instrument Classification for in Situ Rock and Layer Analysis.

K. L. Wagstaff, D. R. Thompson, W. Abbey, A. Allwood, D. Bekker, N. A. Cabrol, Thomas J. Fuchs and K. Ortega.

Geophysical Research Letters, vol. 40, 16, p. 4188–4193, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

TMARKER: A Robust and Free Software Toolkit for Histopathological Cell Counting and Immunohistochemical Staining Estimations.

Peter J. Schueffler, Niels Rupp, Cheng S. Ong, Joachim M. Buhmann, Thomas J. Fuchs and Peter J. Wild.

German Society of Pathology 97th Annual Meeting, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Automated Analysis of Tissue Micro-Array Images on the Example of Renal Cell Carcinoma.

Peter J. Schueffler, Thomas J. Fuchs, Cheng Soon Ong, Volker Roth and Joachim M. Buhmann.

In: Similarity-Based Pattern Analysis and Recognition, p. 219–246, Advances in Computer Vision and Pattern Recognition, Springer, ISBN 978-1-4471-5627-7, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

A Model Development Pipeline for Crohn's Disease Severity Assessment from Magnetic Resonance Images.

Peter J. Schüffler, Dwarikanath Mahapatra, Jeroen A. W. Tielbeek, Franciscus M. Vos, Jesica Makanyanga, Doug A. Pendsé, C. Yung Nio, Jaap Stoker, Stuart A. Taylor and Joachim M. Buhmann.

Outstanding Paper Award

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Feature Selection Strategies for Classifying High Dimensional Astronomical Data Sets.

Ciro Donalek, Arun Kumar A., S. G. Djorgovski, Ashish A. Mahabal, Matthew J. Graham, Thomas J. Fuchs, Michael J. Turmon, N. Sajeeth Philip, Michael Ting-Chang Yang and Giuseppe Longo.

IEEE International Conference on Big Data, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Quickly Boosting Decision Trees – Pruning Underachieving Features using a Provable Bound.

Ron Appel, Piotr Dollar, Thomas J. Fuchs and Pietro Perona.

Proceedings of the 30th International Conference on Machine Learning (ICML), vol. 28, p. 594-602, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Recognizing Humans in Motion: Trajectory-based Areal Video Analysis.

Yumi Iwashita, Michael Ryoo, Thomas J. Fuchs and Curtis Padgett.

24th British Machine Vision Conference (BMVC), 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Automatic Detection and Segmentation of Crohn's Disease Tissues from Abdominal MRI.

D. Mahapatra, P. J. Schüffler, J. A. W. Tielbeek, J. C. Makanyanga, J. Stoker, S. A. Taylor, F. M. Vos and J. M. Buhmann.

IEEE Transactions on Medical Imaging, vol. 32, p. 2332-2347, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

TextureCam: A Smart Camera for Microscale, Mesoscale, and Deep Space.

44th Lunar and Planetary Science Conference, p. 2209, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Structure Preserving Embedding of Dissimilarity Data.

Volker Roth, Thomas J. Fuchs, Julia E. Vogt, Sandhya Prabhakaran and Joachim M. Buhmann.

In: Similarity-Based Pattern Analysis and Recognition, p. 157–178, Advances in Computer Vision and Pattern Recognition, Springer, ISBN 978-1-4471-5627-7, 2013

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

p53 Suppresses Type II Endometrial Carcinomas in Mice and Governs Endometrial Tumour Aggressiveness in Humans.

EMBO Molecular Medicine, vol. 4, 8, p. 808-824, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

End-to-End Dexterous Manipulation with Deliberate Interactive Estimation.

Nicolas H. Hudson, Tom Howard, Jeremy Ma, Abhinandan Jain, Max Bajracharya, Steven Myint, Larry Matthies, Paul Backes, Paul Hebert, Thomas J. Fuchs and Joel Burdick.

IEEE International Conference on Robotics and Automation (ICRA), p. 2371-2378, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Smart Cameras for Remote Science Survey.

David R. Thompson, William Abbey, Abigail Allwood, Dmitriy Bekker, Benjamin Bornstein, Nathalie A. Cabrol, Rebecca Castano, Tara Estlin, Thomas J. Fuchs and Kiri L. Wagstaff.

Proceedings of the 10th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS), 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

Serum and Prostate Cancer Tissue Signatures of ERG Rearrangement Derived from Quantitative Analysis of the PTEN Conditional Knockout Mouse Proteome.

Symposium of the German Society for Pathology, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

A High-Throughput Metabolomics Method to Predict High Concentration Cytotoxicity of Drugs from Low Concentration Profiles.

Stephanie Heux, Thomas J. Fuchs, Joachim Buhmann, Nicola Zamboni and Uwe Sauer.

Metabolomics, vol. 8, 3, p. 433-443, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

TMARKER: A User-Friendly Open-Source Assistance for Tma Grading and Cell Counting.

Peter J. Schueffler, Thomas J. Fuchs, Cheng S. Ong, Peter Wild and Joachim M. Buhmann.

Histopathology Image Analysis (HIMA) Workshop at the 15th International Conference on Medical Image Computing and Computer Assisted Intervention MICCAI, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation

A Seven-Marker Signature and Clinical Outcome in Malignant Melanoma: A Large-Scale Tissue-Microarray Study with Two Independent Patient Cohorts.

PLoS ONE, vol. 7, 6, p. e38222, 2012

PDF URL BibTeX Endnote / RIS Abstract

Download BibTeX citation

Download Endnote/RIS citation