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

Peter J. Schüffler, Judy Sarungbam, Hassan Muhammad, Ed Reznik, Satish Tickoo, Thomas Fuchs
Proceedings of the 1st Machine Learning for Healthcare Conference, PMLR 56:191-208, 2016.

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 alterations to mitochondria between subtypes make 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 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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v56-Schuffler16, title = {Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations}, author = {Schüffler, Peter J. and Sarungbam, Judy and Muhammad, Hassan and Reznik, Ed and Tickoo, Satish and Fuchs, Thomas}, booktitle = {Proceedings of the 1st Machine Learning for Healthcare Conference}, pages = {191--208}, year = {2016}, editor = {Doshi-Velez, Finale and Fackler, Jim and Kale, David and Wallace, Byron and Wiens, Jenna}, volume = {56}, series = {Proceedings of Machine Learning Research}, address = {Northeastern University, Boston, MA, USA}, month = {18--19 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v56/Schuffler16.pdf}, url = {https://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 alterations to mitochondria between subtypes make 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 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.} }
Endnote
%0 Conference Paper %T Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations %A Peter J. Schüffler %A Judy Sarungbam %A Hassan Muhammad %A Ed Reznik %A Satish Tickoo %A Thomas Fuchs %B Proceedings of the 1st Machine Learning for Healthcare Conference %C Proceedings of Machine Learning Research %D 2016 %E Finale Doshi-Velez %E Jim Fackler %E David Kale %E Byron Wallace %E Jenna Wiens %F pmlr-v56-Schuffler16 %I PMLR %P 191--208 %U https://proceedings.mlr.press/v56/Schuffler16.html %V 56 %X 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 alterations to mitochondria between subtypes make 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 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.
RIS
TY - CPAPER TI - Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations AU - Peter J. Schüffler AU - Judy Sarungbam AU - Hassan Muhammad AU - Ed Reznik AU - Satish Tickoo AU - Thomas Fuchs BT - Proceedings of the 1st Machine Learning for Healthcare Conference DA - 2016/12/10 ED - Finale Doshi-Velez ED - Jim Fackler ED - David Kale ED - Byron Wallace ED - Jenna Wiens ID - pmlr-v56-Schuffler16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 56 SP - 191 EP - 208 L1 - http://proceedings.mlr.press/v56/Schuffler16.pdf UR - https://proceedings.mlr.press/v56/Schuffler16.html 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 alterations to mitochondria between subtypes make 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 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. ER -
APA
Schüffler, P.J., Sarungbam, J., Muhammad, H., Reznik, E., Tickoo, S. & Fuchs, T.. (2016). Mitochondria-based Renal Cell Carcinoma Subtyping: Learning from Deep vs. Flat Feature Representations. Proceedings of the 1st Machine Learning for Healthcare Conference, in Proceedings of Machine Learning Research 56:191-208 Available from https://proceedings.mlr.press/v56/Schuffler16.html.

Related Material