Stain-Transforming Cycle-Consistent Generative Adversarial Networks for Improved Segmentation of Renal Histopathology

Thomas de Bel, Meyke Hermsen, Jesper Kers, Jeroen van der Laak, Geert Litjens
Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, PMLR 102:151-163, 2019.

Abstract

The performance of deep learning applications in digital histopathology can deteriorate significantly due to staining variations across centers. We employ cycle-consistent generative adversarial networks (cycleGANs) for unpaired image-to-image translation, facilitating between-center stain transformation. We find that modifications to the original cycleGAN architecture make it more suitable for stain transformation, creating artificially stained images of high quality. Specifically, changing the generator model to a smaller U-net-like architecture, adding an identity loss term, increasing the batch size and the learning all led to improved training stability and performance. Furthermore, we propose a method for dealing with tiling artifacts when applying the network on whole slide images (WSIs). We apply our stain transformation method on two datasets of PAS-stained (Periodic Acid-Schiff) renal tissue sections from different centers. We show that stain transformation is beneficial to the performance of cross-center segmentation, raising the Dice coefficient from 0.36 to 0.85 and from 0.45 to 0.73 on the two datasets.

Cite this Paper


BibTeX
@InProceedings{pmlr-v102-de-bel19a, title = {Stain-Transforming Cycle-Consistent Generative Adversarial Networks for Improved Segmentation of Renal Histopathology}, author = {{de Bel}, Thomas and Hermsen, Meyke and Kers, Jesper and {van der Laak}, Jeroen and Litjens, Geert}, booktitle = {Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning}, pages = {151--163}, year = {2019}, editor = {Cardoso, M. Jorge and Feragen, Aasa and Glocker, Ben and Konukoglu, Ender and Oguz, Ipek and Unal, Gozde and Vercauteren, Tom}, volume = {102}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v102/de-bel19a/de-bel19a.pdf}, url = {https://proceedings.mlr.press/v102/de-bel19a.html}, abstract = {The performance of deep learning applications in digital histopathology can deteriorate significantly due to staining variations across centers. We employ cycle-consistent generative adversarial networks (cycleGANs) for unpaired image-to-image translation, facilitating between-center stain transformation. We find that modifications to the original cycleGAN architecture make it more suitable for stain transformation, creating artificially stained images of high quality. Specifically, changing the generator model to a smaller U-net-like architecture, adding an identity loss term, increasing the batch size and the learning all led to improved training stability and performance. Furthermore, we propose a method for dealing with tiling artifacts when applying the network on whole slide images (WSIs). We apply our stain transformation method on two datasets of PAS-stained (Periodic Acid-Schiff) renal tissue sections from different centers. We show that stain transformation is beneficial to the performance of cross-center segmentation, raising the Dice coefficient from 0.36 to 0.85 and from 0.45 to 0.73 on the two datasets.} }
Endnote
%0 Conference Paper %T Stain-Transforming Cycle-Consistent Generative Adversarial Networks for Improved Segmentation of Renal Histopathology %A Thomas de Bel %A Meyke Hermsen %A Jesper Kers %A Jeroen van der Laak %A Geert Litjens %B Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2019 %E M. Jorge Cardoso %E Aasa Feragen %E Ben Glocker %E Ender Konukoglu %E Ipek Oguz %E Gozde Unal %E Tom Vercauteren %F pmlr-v102-de-bel19a %I PMLR %P 151--163 %U https://proceedings.mlr.press/v102/de-bel19a.html %V 102 %X The performance of deep learning applications in digital histopathology can deteriorate significantly due to staining variations across centers. We employ cycle-consistent generative adversarial networks (cycleGANs) for unpaired image-to-image translation, facilitating between-center stain transformation. We find that modifications to the original cycleGAN architecture make it more suitable for stain transformation, creating artificially stained images of high quality. Specifically, changing the generator model to a smaller U-net-like architecture, adding an identity loss term, increasing the batch size and the learning all led to improved training stability and performance. Furthermore, we propose a method for dealing with tiling artifacts when applying the network on whole slide images (WSIs). We apply our stain transformation method on two datasets of PAS-stained (Periodic Acid-Schiff) renal tissue sections from different centers. We show that stain transformation is beneficial to the performance of cross-center segmentation, raising the Dice coefficient from 0.36 to 0.85 and from 0.45 to 0.73 on the two datasets.
APA
de Bel, T., Hermsen, M., Kers, J., van der Laak, J. & Litjens, G.. (2019). Stain-Transforming Cycle-Consistent Generative Adversarial Networks for Improved Segmentation of Renal Histopathology. Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 102:151-163 Available from https://proceedings.mlr.press/v102/de-bel19a.html.

Related Material