Diffusion Models with Implicit Guidance for Medical Anomaly Detection

TL;DR

This paper presents THOR, a novel diffusion model approach with implicit guidance that enhances anomaly detection accuracy in medical images by better preserving healthy tissue during image restoration.

Contribution

The paper introduces THOR, a new method that incorporates implicit guidance via temporal anomaly maps into diffusion models for improved medical anomaly detection.

Findings

THOR outperforms existing diffusion methods in brain MRI anomaly detection.

THOR achieves higher segmentation accuracy and fewer false positives.

The approach effectively preserves healthy tissue during image restoration.

Abstract

Diffusion models have advanced unsupervised anomaly detection by improving the transformation of pathological images into pseudo-healthy equivalents. Nonetheless, standard approaches may compromise critical information during pathology removal, leading to restorations that do not align with unaffected regions in the original scans. Such discrepancies can inadvertently increase false positive rates and reduce specificity, complicating radiological evaluations. This paper introduces Temporal Harmonization for Optimal Restoration (THOR), which refines the de-noising process by integrating implicit guidance through temporal anomaly maps. THOR aims to preserve the integrity of healthy tissue in areas unaffected by pathology. Comparative evaluations show that THOR surpasses existing diffusion-based methods in detecting and segmenting anomalies in brain MRIs and wrist X-rays. Code: https://github.com/ci-ber/THOR_DDPM.