Structural-Functional Connectome Generation via Diffusion-Guided Graph Transformer for Alzheimer's Disease Analysis.

Area of Science:

• Neuroscience
• Medical Imaging
• Computational Biology

Background:

• Alzheimer's Disease (AD) diagnosis relies on understanding brain network abnormalities.
• Current methods analyze structural or functional networks separately, missing crucial complementary information.
• Existing tools are time-consuming and require subjective parameter tuning.

Purpose of the Study:

• To develop a novel paradigm, DiffusionBrain, for generating integrated structural-functional brain networks.
• To apply DiffusionBrain for enhanced diagnosis and analysis of Alzheimer's Disease.
• To overcome limitations of unimodal brain network analysis.

Main Methods:

• Designed a Graph Prompt Fusion Module (GPFM) for multimodal feature extraction and fusion.
• Developed a Dual Diffusion-guided Graph Transformer (DDGT) for efficient network generation.
• Implemented a Graph Alignment Module (GAM) for deep fusion of structural and functional networks.
• Utilized the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset for validation.

Main Results:

• DiffusionBrain effectively captures complex interactions between structural and functional brain networks.
• The method successfully reveals abnormal connection patterns characteristic of Alzheimer's Disease.
• Demonstrated superior performance in modeling multimodal brain networks compared to existing approaches.

Conclusions:

• DiffusionBrain offers a powerful new paradigm for multimodal brain network modeling.
• This approach enhances the identification of Alzheimer's Disease biomarkers and pathological mechanisms.
• Provides critical insights for early diagnosis and intervention strategies in Alzheimer's Disease.