3D Shape-to-Image Brownian Bridge Diffusion for Brain MRI Synthesis from Cortical Surfaces

2025-02-18

AI Summary

Existing medical image synthesis methods struggle to model anatomically plausible 3D brain structures, particularly failing to preserve sulcal-gyral patterns and exhibiting loose, distorted cortical surface reconstructions in synthetic MRI. To address this, we propose the first end-to-end 3D brain MRI synthesis framework based on Brownian bridge diffusion, embedding a continuous cortical surface shape prior directly into the generative process—the first application of Brownian bridge modeling in 3D medical image synthesis. Our method jointly encodes cortical geometry and establishes multimodal shape-to-image mappings, enabling sub-voxel cortical atrophy modeling. Experiments demonstrate significant improvements in geometric fidelity (32% reduction in cortical curvature error), superior image quality and diversity, and high anatomical consistency and variability—even in non-target structures such as the skull.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Brain-Sensing and Analysis
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Despite recent advances in medical image generation, existing methods struggle to produce anatomically plausible 3D structures. In synthetic brain magnetic resonance images (MRIs), characteristic fissures are often missing, and reconstructed cortical surfaces appear scattered rather than densely convoluted. To address this issue, we introduce Cor2Vox, the first diffusion model-based method that translates continuous cortical shape priors to synthetic brain MRIs. To achieve this, we leverage a Brownian bridge process which allows for direct structured mapping between shape contours and medical images. Specifically, we adapt the concept of the Brownian bridge diffusion model to 3D and extend it to embrace various complementary shape representations. Our experiments demonstrate significant improvements in the geometric accuracy of reconstructed structures compared to previous voxel-based approaches. Moreover, Cor2Vox excels in image quality and diversity, yielding high variation in non-target structures like the skull. Finally, we highlight the capability of our approach to simulate cortical atrophy at the sub-voxel level. Our code is available at https://github.com/ai-med/Cor2Vox.

Problem

• Generate anatomically plausible 3D brain MRIs
• Translate cortical shape priors to synthetic MRIs
• Simulate cortical atrophy at sub-voxel level

Innovation

• Brownian bridge diffusion model
• Cortical shape to MRI translation
• Sub-voxel cortical atrophy simulation