Unsupervised Anomaly Detection in Brain MRI via Disentangled Anatomy Learning

2025-12-26

AI Summary

To address poor generalizability and persistent abnormality leakage in pseudo-healthy image (PHI) reconstruction within unsupervised anomaly detection for multi-center, multi-modal brain MRI, this paper proposes a reconstruction framework that disentangles anatomical structure from imaging information. Our method comprises two key innovations: (1) a disentangled representation module leveraging anatomical priors and differentiable one-hot encoding to achieve imaging-invariant anatomical modeling; and (2) an edge-driven reconstruction module integrating high-frequency edge features with a disentangle-then-recombine architecture to suppress abnormality propagation. We conduct systematic evaluation across nine public datasets (4,443 scans), demonstrating consistent superiority over 17 state-of-the-art methods—achieving +18.32% average precision (AP) and +13.64% Dice Similarity Coefficient (DSC). The framework significantly enhances cross-center and cross-modality generalizability.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Detection of various lesions in brain MRI is clinically critical, but challenging due to the diversity of lesions and variability in imaging conditions. Current unsupervised learning methods detect anomalies mainly through reconstructing abnormal images into pseudo-healthy images (PHIs) by normal samples learning and then analyzing differences between images. However, these unsupervised models face two significant limitations: restricted generalizability to multi-modality and multi-center MRIs due to their reliance on the specific imaging information in normal training data, and constrained performance due to abnormal residuals propagated from input images to reconstructed PHIs. To address these limitations, two novel modules are proposed, forming a new PHI reconstruction framework. Firstly, the disentangled representation module is proposed to improve generalizability by decoupling brain MRI into imaging information and essential imaging-invariant anatomical images, ensuring that the reconstruction focuses on the anatomy. Specifically, brain anatomical priors and a differentiable one-hot encoding operator are introduced to constrain the disentanglement results and enhance the disentanglement stability. Secondly, the edge-to-image restoration module is designed to reconstruct high-quality PHIs by restoring the anatomical representation from the high-frequency edge information of anatomical images, and then recoupling the disentangled imaging information. This module not only suppresses abnormal residuals in PHI by reducing abnormal pixels input through edge-only input, but also effectively reconstructs normal regions using the preserved structural details in the edges. Evaluated on nine public datasets (4,443 patients' MRIs from multiple centers), our method outperforms 17 SOTA methods, achieving absolute improvements of +18.32% in AP and +13.64% in DSC.

Problem

• Improves generalizability of anomaly detection across multi-modality and multi-center brain MRIs
• Reduces abnormal residuals in pseudo-healthy image reconstruction via edge-based restoration
• Enhances disentanglement of imaging and anatomical information for more accurate lesion detection

Innovation

• Disentangled representation module decouples MRI into imaging and anatomy
• Edge-to-image restoration reconstructs healthy images from anatomical edges
• Method improves generalizability across multi-modality and multi-center MRI datasets