CRADL: Contrastive Representations for Unsupervised Anomaly Detection and Localization

TL;DR

CRADL introduces a contrastive learning-based framework for unsupervised anomaly detection and localization in medical imaging, emphasizing semantic-rich representations over traditional generative models.

Contribution

The paper proposes CRADL, a novel approach that models normal data in a contrastive representation space, improving anomaly detection and localization performance.

Findings

Contrastive representations outperform generative latent models.

CRADL achieves competitive or superior results on multiple datasets.

Semantic-rich features enhance anomaly detection accuracy.

Abstract

Unsupervised anomaly detection in medical imaging aims to detect and localize arbitrary anomalies without requiring annotated anomalous data during training. Often, this is achieved by learning a data distribution of normal samples and detecting anomalies as regions in the image which deviate from this distribution. Most current state-of-the-art methods use latent variable generative models operating directly on the images. However, generative models have been shown to mostly capture low-level features, s.a. pixel-intensities, instead of rich semantic features, which also applies to their representations. We circumvent this problem by proposing CRADL whose core idea is to model the distribution of normal samples directly in the low-dimensional representation space of an encoder trained with a contrastive pretext-task. By utilizing the representations of contrastive learning, we aim to fix the over-fixation on low-level features and learn more semantic-rich representations. Our experiments on anomaly detection and localization tasks using three distinct evaluation datasets show that 1) contrastive representations are superior to representations of generative latent variable models and 2) the CRADL framework shows competitive or superior performance to state-of-the-art.