Lesion-DDPM: Lesion-Enhanced 3D Diffusion for MS MRI Synthesis

2026-06-13

AI Summary

This work addresses the challenge of synthesizing realistic, sparse multiple sclerosis (MS) lesions in brain MRI, which is hindered by data scarcity and heterogeneity. To this end, the authors propose a lesion-aware 3D conditional diffusion model that uniquely integrates lesion-weighted reconstruction loss with multi-scale anatomical mask injection. Built upon a 3D DDPM architecture, the method preserves global brain anatomy while significantly enhancing lesion fidelity. Experimental results demonstrate superior lesion reconstruction accuracy over existing approaches. Notably, a 3D U-Net trained solely on synthetic data achieves a Dice score of 0.616 for lesion segmentation, which further improves to 0.685 when combined with real data, thereby effectively boosting downstream segmentation performance.

Abstract

3D FLAIR MRI is widely recommended as one of the standard MRI sequences for brain imaging in multiple sclerosis (MS), but publicly available MS datasets remain relatively small and vary across scanners, acquisition protocols, and lesion patterns. This scarcity and variability hinder the development of robust neuroimaging machine learning models and are particularly challenging for generative models that aim to synthesize images while preserving small, sparse lesions. We propose Lesion-DDPM, a 3D conditional diffusion framework for lesion-aware FLAIR synthesis that incorporates multi-level anatomical mask injection together with a lesion-weighted reconstruction loss to emphasize lesion voxels while maintaining global brain structure. Using a curated subset of the MSLesSeg dataset, we compare Lesion-DDPM with representative state-of-the-art GAN- and diffusion-based models, assessing both image-generation metrics and downstream 3D U-Net segmentation. In our experiments, Lesion-DDPM achieved the lowest lesion-region reconstruction error among all methods. In a downstream 3D U-Net lesion segmentation task, a model trained only on Lesion-DDPM-generated scans and evaluated on real MRIs reached a Dice score of 0.616 compared with 0.569 for the best competing synthetic dataset. When Lesion-DDPM images were added to the real training set, the Dice score further increased to 0.685.

Problem

• MS MRI
• lesion preservation
• data scarcity
• image synthesis
• scanner variability

Innovation

• Lesion-DDPM
• 3D diffusion model
• lesion-aware synthesis
• FLAIR MRI
• lesion-weighted loss