Medical Image Anomaly Detection

Overview

Medical image anomaly detection identifies findings that differ from expected anatomy, supporting screening, quality assurance, and discovery of rare or previously unseen disease. Common approaches learn normal patterns from healthy scans using reconstruction, feature-distance, density, or prototype models, then flag and localize deviations; other methods use synthetic lesions or small labeled examples. Foundation models and vision-language systems increasingly enable zero-shot or few-shot detection across modalities, sometimes adding anatomical context or visual explanations. A central challenge is distinguishing subtle pathology from normal variation while generalizing across scanners, institutions, and modalities, so external evaluation and reliable lesion localization remain important.

Latest papers

CardsList

94 papers7 papers in the last four weeks, down 22% on the four weeks before. 0.1% of all new papers.

  1. Toxicity Assessment in Preclinical Histopathology via Class-Aware Mahalanobis Distance for Known and Novel Anomalies

    Feb 2, 2026Olga Graf, Dhrupal Patel, Peter Groß +3Histopathology Image SegmentationMahalanobis Distance

  2. MedAD-R1: Consistency-Reinforced Policy Optimization for Interpretable Medical Anomaly Detection

    Feb 1, 2026Haitao Zhang, Yingying Wang, Jiaxiang Wang +5HealthcareGroup Relative Policy Optimization

  3. Diverse Normal Prototypes-Guided Contrastive Reconstruction for Medical Anomaly Detection

    Aug 27, 2025Luhu Li, Bin Liu, Bowen Lin +3Prototype LearningMedical Image Anomaly Detection