DeScarGAN: Disease-Specific Anomaly Detection with Weak Supervision

TL;DR

DeScarGAN is a weakly supervised, detail-preserving method for detecting disease-specific structural anomalies in medical images, outperforming existing methods on synthetic and chest X-ray datasets.

Contribution

It introduces a novel weakly supervised approach that leverages disease-specific information from patient and control groups for detailed anomaly detection.

Findings

Outperforms state-of-the-art methods on synthetic data

Shows superior detection in chest X-ray images

Effectively detects structural changes in medical images

Abstract

Anomaly detection and localization in medical images is a challenging task, especially when the anomaly exhibits a change of existing structures, e.g., brain atrophy or changes in the pleural space due to pleural effusions. In this work, we present a weakly supervised and detail-preserving method that is able to detect structural changes of existing anatomical structures. In contrast to standard anomaly detection methods, our method extracts information about the disease characteristics from two groups: a group of patients affected by the same disease and a healthy control group. Together with identity-preserving mechanisms, this enables our method to extract highly disease-specific characteristics for a more detailed detection of structural changes. We designed a specific synthetic data set to evaluate and compare our method against state-of-the-art anomaly detection methods. Finally, we show the performance of our method on chest X-ray images. Our method called DeScarGAN outperforms other anomaly detection methods on the synthetic data set and by visual inspection on the chest X-ray image data set.