Self-supervised Out-of-distribution Detection for Cardiac CMR Segmentation

Camila Gonzalez, Anirban Mukhopadhyay
Proceedings of the Fourth Conference on Medical Imaging with Deep Learning, PMLR 143:205-218, 2021.

Abstract

The segmentation of cardiac structures in Cine Magnetic Resonance imaging (CMR) plays an important role in monitoring ventricular function, and many deep learning solutions have been introduced that successfully automate this task. Yet due to variabilities in the CMR acquisition process, images from different centers or acquisition protocols differ considerably. This causes deep learning models to fail silently. It is therefore crucial to identify out-of-distribution (OOD) samples for which the trained model is unsuitable. For models with a self-supervised proxy task, we propose a simple method to identify OOD samples that does not require adapting the model architecture or access to a separate OOD dataset during training. As the performance of self-supervised tasks can be assessed without ground truth information, it indicates during test time when a sample differs from the training distribution. The proposed method combines a voxel-wise uncertainty estimate with the self-supervision information. Our approach is validated across three CMR datasets and two different proxy tasks. We find that it is more effective at detecting OOD samples than state-of-the-art post-hoc OOD detection and uncertainty estimation approaches.

Cite this Paper

BibTeX
@InProceedings{pmlr-v143-gonzalez21a, title = {Self-supervised Out-of-distribution Detection for Cardiac {CMR} Segmentation}, author = {Gonzalez, Camila and Mukhopadhyay, Anirban}, booktitle = {Proceedings of the Fourth Conference on Medical Imaging with Deep Learning}, pages = {205--218}, year = {2021}, editor = {Heinrich, Mattias and Dou, Qi and de Bruijne, Marleen and Lellmann, Jan and Schläfer, Alexander and Ernst, Floris}, volume = {143}, series = {Proceedings of Machine Learning Research}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v143/gonzalez21a/gonzalez21a.pdf}, url = {https://proceedings.mlr.press/v143/gonzalez21a.html}, abstract = {The segmentation of cardiac structures in Cine Magnetic Resonance imaging (CMR) plays an important role in monitoring ventricular function, and many deep learning solutions have been introduced that successfully automate this task. Yet due to variabilities in the CMR acquisition process, images from different centers or acquisition protocols differ considerably. This causes deep learning models to fail silently. It is therefore crucial to identify out-of-distribution (OOD) samples for which the trained model is unsuitable. For models with a self-supervised proxy task, we propose a simple method to identify OOD samples that does not require adapting the model architecture or access to a separate OOD dataset during training. As the performance of self-supervised tasks can be assessed without ground truth information, it indicates during test time when a sample differs from the training distribution. The proposed method combines a voxel-wise uncertainty estimate with the self-supervision information. Our approach is validated across three CMR datasets and two different proxy tasks. We find that it is more effective at detecting OOD samples than state-of-the-art post-hoc OOD detection and uncertainty estimation approaches.} }
Endnote
%0 Conference Paper %T Self-supervised Out-of-distribution Detection for Cardiac CMR Segmentation %A Camila Gonzalez %A Anirban Mukhopadhyay %B Proceedings of the Fourth Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2021 %E Mattias Heinrich %E Qi Dou %E Marleen de Bruijne %E Jan Lellmann %E Alexander Schläfer %E Floris Ernst %F pmlr-v143-gonzalez21a %I PMLR %P 205--218 %U https://proceedings.mlr.press/v143/gonzalez21a.html %V 143 %X The segmentation of cardiac structures in Cine Magnetic Resonance imaging (CMR) plays an important role in monitoring ventricular function, and many deep learning solutions have been introduced that successfully automate this task. Yet due to variabilities in the CMR acquisition process, images from different centers or acquisition protocols differ considerably. This causes deep learning models to fail silently. It is therefore crucial to identify out-of-distribution (OOD) samples for which the trained model is unsuitable. For models with a self-supervised proxy task, we propose a simple method to identify OOD samples that does not require adapting the model architecture or access to a separate OOD dataset during training. As the performance of self-supervised tasks can be assessed without ground truth information, it indicates during test time when a sample differs from the training distribution. The proposed method combines a voxel-wise uncertainty estimate with the self-supervision information. Our approach is validated across three CMR datasets and two different proxy tasks. We find that it is more effective at detecting OOD samples than state-of-the-art post-hoc OOD detection and uncertainty estimation approaches.
APA
Gonzalez, C. & Mukhopadhyay, A.. (2021). Self-supervised Out-of-distribution Detection for Cardiac CMR Segmentation. Proceedings of the Fourth Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 143:205-218 Available from https://proceedings.mlr.press/v143/gonzalez21a.html.

Related Material