Unsupervised Anomaly Detection for Image Dataset Quality Assurance in Multi-Center Breast MRI.

arXiv:2608.16725v2 Announce Type: replace-cross Abstract: Corrupted, inconsistent, or anomalous data silently threatens the safety and reliability of medical AI. Despite growing regulatory recognition of dataset quality assurance (QA) for high-risk medical AI,