“…Recently, a wide range of diffusion-based perception applications has emerged, such as image generation [15,21,35,47], image segmentation [4,7,18,23], object detection [8], etc. A few previous works have tentatively explored the application of diffusion in medical anomaly detection [32,53], such as AnoDDPM [21], but its inference speed is slow and the anomalous false positive rate is high, which is nowhere near the requirements for practical deployment. To the best of our knowledge, we are the first to propose a diffusion-based pipeline with gratifying anomaly detection and localization performance as well as fast inference speed.…”