A universal 'one-shot' medical anomaly detector that outperforms specialized models across nine different datasets.
March 26, 2026
Original Paper
Semantic Iterative Reconstruction: One-Shot Universal Anomaly Detection
arXiv · 2603.23766
The Takeaway
Current medical AI usually requires hundreds of 'normal' images for every new organ or modality. This framework uses a single model and exactly one normal sample from diverse domains to detect anomalies, representing a major step toward general-purpose, clinical-grade diagnostic tools.
From the abstract
Unsupervised medical anomaly detection is severely limited by the scarcity of normal training samples. Existing methods typically train dedicated models for each dataset or disease, requiring hundreds of normal images per task and lacking cross-modality generalization.We propose Semantic Iterative Reconstruction (SIR), a framework that enables a single universal model to detect anomalies across diverse medical domains using extremely few normal samples. SIR leverages a pretrained teacher encoder