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

Supervised deep learning techniques show promise in medical image analysis.However, they require comprehensive annotated data sets, which poseschallenges, particularly for rare diseases. Consequently, unsupervised anomalydetection (UAD) emerges as a viable alternative for pathology segmentation, asonly healthy data is required for training. However, recent UAD anomaly scoringfunctions often focus on intensity only and neglect structural differences,which impedes the segmentation performance. This work investigates thepotential of Structural Similarity (SSIM) to bridge this gap. SSIM capturesboth intensity and structural disparities and can be advantageous over theclassical l1 error. However, we show that there is more than one optimalkernel size for the SSIM calculation for different pathologies. Therefore, weinvestigate an adaptive ensembling strategy for various kernel sizes to offer amore pathology-agnostic scoring mechanism. We demonstrate that this ensemblingstrategy can enhance the performance of DMs and mitigate the sensitivity todifferent kernel sizes across varying pathologies, highlighting its promise forbrain MRI anomaly detection.