Unsupervised learning can leverage large-scale data sources without the need for annotations. In this context, deep learning-based auto encoders have shown great potential in detecting anomalies in medical images. However, state-of-the-art anomaly scores are still based on the reconstruction error, which lacks in two essential parts: it ignores the model-internal representation employed for reconstruction, and it lacks formal assertions and comparability between samples. We address these shortcomings by proposing the Context-encoding Variational Autoencoder (ceVAE) which combines reconstruction- with density-based anomaly scoring. This improves the sample- as well as pixel-wise results. In our experiments on the BraTS-2017 and ISLES-2015 segmentation benchmarks, the ceVAE achieves unsupervised ROC-AUCs of 0.95 and 0.89, respectively, thus outperforming state-of-the-art methods by a considerable margin.
@article{arxiv.1812.05941,
title = {Context-encoding Variational Autoencoder for Unsupervised Anomaly Detection},
author = {David Zimmerer and Simon A. A. Kohl and Jens Petersen and Fabian Isensee and Klaus H. Maier-Hein},
journal= {arXiv preprint arXiv:1812.05941},
year = {2018}
}