A Multi-Source Ultrasound Benchmark Revealing the Limits of Contemporary Self-Supervised Anomaly Detection Methods

Key points

• Self-supervised anomaly detection is a promising paradigm for medical ultrasound, as normal images are often easier to obtain than exhaustive annotations of all possible pathologies.1
• The goal of SADUSI is to provide a diverse normal ultrasound distribution and a benchmark for visible structural anomalies that can be assessed from single images.1
• In particular, reconstruction-based diffusion methods such as AnoDDPM and DeCo-Diff achieve pixel-level AUROC values of 0.56-0.72 and maximum F1 scores of 0.10-0.26, indicating limited separation of pathology from normal image regions.1
• These findings suggest that broad multi-source ultrasound anomaly detection remains an open challenge and that SADUSI can serve as a resource for developing methods that generalize beyond anatomy-specific settings.1

Sources (1)

• [1]A Multi-Source Ultrasound Benchmark Revealing the Limits of Contemporary Self-Supervised Anomaly Detection Methods
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 08:39 AM

  We introduce the SADUSI benchmark, a multi-source ultrasound dataset designed to train and evaluate anomaly detection methods across a broad range of anatomical regions, views, and acquisition protocols.

  Self-supervised anomaly detection is a promising paradigm for medical ultrasound, as normal images are often easier to obtain than exhaustive annotations of all possible pathologies.

Extractive summary: sentences quoted from the sources.