DNP-Guided Contrastive Reconstruction with a Reverse Distillation Transformer for Medical Anomaly Detection

TL;DR

This paper introduces a novel framework for medical anomaly detection that combines a trainable encoder, prototype-guided reconstruction, and a diversity-aware loss to improve localization accuracy and interpretability in limited annotation settings.

Contribution

It proposes a unified approach integrating a momentum-enhanced trainable encoder with prototype-guided reconstruction and a new diversity-aware loss to prevent prototype collapse and enhance domain adaptation.

Findings

Outperforms prior methods on multiple medical imaging benchmarks.

Improves anomaly localization accuracy and interpretability.

Effectively prevents prototype collapse through diversity constraints.

Abstract

Anomaly detection in medical images is challenging due to limited annotations and a domain gap compared to natural images. Existing reconstruction methods often rely on frozen pre-trained encoders, which limits adaptation to domain-specific features and reduces localization accuracy. Prototype-based learning offers interpretability and clustering benefits but suffers from prototype collapse, where few prototypes dominate training, harming diversity and generalization. To address this, we propose a unified framework combining a trainable encoder with prototype-guided reconstruction and a novel Diversity-Aware Alignment Loss. The trainable encoder, enhanced by a momentum branch, enables stable domain-adaptive feature learning. A lightweight Prototype Extractor mines informative normal prototypes to guide the decoder via attention for precise reconstruction. Our loss enforces balanced prototype use through diversity constraints and per-prototype normalization, effectively preventing collapse. Experiments on multiple medical imaging benchmarks show significant improvements in representation quality and anomaly localization, outperforming prior methods. Visualizations and prototype assignment analyses further validate the effectiveness of our anti-collapse mechanism and enhanced interpretability.