Diffusion Models with Ensembled Structure-Based Anomaly Scoring for Unsupervised Anomaly Detection

arXiv:2403.14262v1 Announce Type: cross
Abstract: Supervised deep learning techniques show promise in medical image analysis. However, they require comprehensive annotated data sets, which poses challenges, particularly for rare diseases. Consequently, unsupervised anomaly detection (UAD) emerges as a viable alternative for pathology segmentation, as only healthy data is required for training. However, recent UAD anomaly scoring functions often focus on intensity only and neglect structural differences, which impedes the segmentation performance. This work investigates the potential of Structural Similarity (SSIM) to …

abstract analysis annotated data anomaly anomaly detection arxiv challenges cs.cv cs.lg data data sets deep learning deep learning techniques detection diffusion diffusion models diseases eess.iv however image medical pathology rare diseases scoring segmentation show type unsupervised