MAD-AD: Masked Diffusion for Unsupervised Brain Anomaly Detection

TL;DR

MAD-AD introduces a novel masked diffusion approach that learns normal brain anatomy representations to accurately detect and localize anomalies in MRI scans without requiring labeled abnormal data.

Contribution

It proposes a masking diffusion model that effectively captures normal brain structures and isolates anomalies as noise, improving unsupervised anomaly detection in medical imaging.

Findings

Outperforms existing unsupervised methods in anomaly localization

Generates accurate normal counterparts for anomalous patches

Effectively isolates anomalies as noise in latent space

Abstract

Unsupervised anomaly detection in brain images is crucial for identifying injuries and pathologies without access to labels. However, the accurate localization of anomalies in medical images remains challenging due to the inherent complexity and variability of brain structures and the scarcity of annotated abnormal data. To address this challenge, we propose a novel approach that incorporates masking within diffusion models, leveraging their generative capabilities to learn robust representations of normal brain anatomy. During training, our model processes only normal brain MRI scans and performs a forward diffusion process in the latent space that adds noise to the features of randomly-selected patches. Following a dual objective, the model learns to identify which patches are noisy and recover their original features. This strategy ensures that the model captures intricate patterns of normal brain structures while isolating potential anomalies as noise in the latent space. At inference, the model identifies noisy patches corresponding to anomalies and generates a normal counterpart for these patches by applying a reverse diffusion process. Our method surpasses existing unsupervised anomaly detection techniques, demonstrating superior performance in generating accurate normal counterparts and localizing anomalies. The code is available at hhttps://github.com/farzad-bz/MAD-AD.