An Auto-Encoder Strategy for Adaptive Image Segmentation

Evan M. Yu, Juan Eugenio Iglesias, Adrian V. Dalca, Mert R. Sabuncu
Proceedings of the Third Conference on Medical Imaging with Deep Learning, PMLR 121:881-891, 2020.

Abstract

Deep neural networks are powerful tools for biomedical image segmentation. These models are often trained with heavy supervision, relying on pairs of images and corresponding voxel-level labels. However, obtaining segmentations of anatomical regions on a large number of cases can be prohibitively expensive. Thus there is a strong need for deep learning-based segmentation tools that do not require heavy supervision and can continuously adapt. In this paper, we propose a novel perspective of segmentation as a discrete representation learning problem, and present a variational autoencoder segmentation strategy that is flexible and adaptive. Our method, called Segmentation Auto-Encoder (SAE), leverages all available unlabeled scans and merely requires a segmentation prior, which can be \textit{a single unpaired} segmentation image. In experiments, we apply SAE to brain MRI scans. Our results show that SAE can produce good quality segmentations, particularly when the prior is good. We demonstrate that a Markov Random Field prior can yield significantly better results than a spatially independent prior. Our code is freely available at: {https://github.com/evanmy/sae}.

Cite this Paper


BibTeX
@InProceedings{pmlr-v121-yu20a, title = {An Auto-Encoder Strategy for Adaptive Image Segmentation}, author = {Yu, Evan M. and Iglesias, Juan Eugenio and Dalca, Adrian V. and Sabuncu, Mert R.}, booktitle = {Proceedings of the Third Conference on Medical Imaging with Deep Learning}, pages = {881--891}, year = {2020}, editor = {Arbel, Tal and Ben Ayed, Ismail and de Bruijne, Marleen and Descoteaux, Maxime and Lombaert, Herve and Pal, Christopher}, volume = {121}, series = {Proceedings of Machine Learning Research}, month = {06--08 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v121/yu20a/yu20a.pdf}, url = {https://proceedings.mlr.press/v121/yu20a.html}, abstract = {Deep neural networks are powerful tools for biomedical image segmentation. These models are often trained with heavy supervision, relying on pairs of images and corresponding voxel-level labels. However, obtaining segmentations of anatomical regions on a large number of cases can be prohibitively expensive. Thus there is a strong need for deep learning-based segmentation tools that do not require heavy supervision and can continuously adapt. In this paper, we propose a novel perspective of segmentation as a discrete representation learning problem, and present a variational autoencoder segmentation strategy that is flexible and adaptive. Our method, called Segmentation Auto-Encoder (SAE), leverages all available unlabeled scans and merely requires a segmentation prior, which can be \textit{a single unpaired} segmentation image. In experiments, we apply SAE to brain MRI scans. Our results show that SAE can produce good quality segmentations, particularly when the prior is good. We demonstrate that a Markov Random Field prior can yield significantly better results than a spatially independent prior. Our code is freely available at: {https://github.com/evanmy/sae}.} }
Endnote
%0 Conference Paper %T An Auto-Encoder Strategy for Adaptive Image Segmentation %A Evan M. Yu %A Juan Eugenio Iglesias %A Adrian V. Dalca %A Mert R. Sabuncu %B Proceedings of the Third Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2020 %E Tal Arbel %E Ismail Ben Ayed %E Marleen de Bruijne %E Maxime Descoteaux %E Herve Lombaert %E Christopher Pal %F pmlr-v121-yu20a %I PMLR %P 881--891 %U https://proceedings.mlr.press/v121/yu20a.html %V 121 %X Deep neural networks are powerful tools for biomedical image segmentation. These models are often trained with heavy supervision, relying on pairs of images and corresponding voxel-level labels. However, obtaining segmentations of anatomical regions on a large number of cases can be prohibitively expensive. Thus there is a strong need for deep learning-based segmentation tools that do not require heavy supervision and can continuously adapt. In this paper, we propose a novel perspective of segmentation as a discrete representation learning problem, and present a variational autoencoder segmentation strategy that is flexible and adaptive. Our method, called Segmentation Auto-Encoder (SAE), leverages all available unlabeled scans and merely requires a segmentation prior, which can be \textit{a single unpaired} segmentation image. In experiments, we apply SAE to brain MRI scans. Our results show that SAE can produce good quality segmentations, particularly when the prior is good. We demonstrate that a Markov Random Field prior can yield significantly better results than a spatially independent prior. Our code is freely available at: {https://github.com/evanmy/sae}.
APA
Yu, E.M., Iglesias, J.E., Dalca, A.V. & Sabuncu, M.R.. (2020). An Auto-Encoder Strategy for Adaptive Image Segmentation. Proceedings of the Third Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 121:881-891 Available from https://proceedings.mlr.press/v121/yu20a.html.

Related Material