Context-encoding Variational Autoencoder for Unsupervised Anomaly Detection

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,...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Zimmerer, David, Kohl, Simon A A, Petersen, Jens, Isensee, Fabian, Maier-Hein, Klaus H
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 14.12.2018
Subjects
Annotations
Anomalies
Coders
Coding
Image detection
Image segmentation
Machine learning
Medical imaging
Reconstruction
State of the art
Online AccessGet full text

Cover

More Information
Summary: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.
ISSN:2331-8422