GPU Accelerated Modeling of Cortical Radial and Tangential Connectivity Changes in Neurodegeneration

Area of Science:

• Neuroimaging
• Neuroscience
• Medical Physics

Background:

• Diffusion MRI signals in the cerebral cortex are linked to neurodegenerative diseases.
• Existing models like Neurite Orientation Dispersion and Density Imaging (NODDI) have limitations in capturing orientation-specific cortical connectivity changes.
• Detailed characterization of cortical microstructure is crucial for understanding neurodegeneration.

Purpose of the Study:

• To develop and validate a novel method for decomposing cortical tissue diffusion signals into radial and tangential components.
• To improve the characterization of orientation-specific connectivity changes within the human cerebral cortex.
• To assess the method's ability to detect cortical gray matter changes associated with tau pathology in Alzheimer's disease.

Main Methods:

• Utilized multi-shell diffusion imaging data combined with anatomical information from brain surfaces.
• Developed a GPU-accelerated probabilistic optimization framework for signal decomposition.
• Estimated radial and tangential diffusion components, ensuring smoothness and anatomical consistency.

Main Results:

• The proposed method successfully decomposed cortical diffusion signals into radial and tangential components.
• Demonstrated enhanced sensitivity in revealing cortical gray matter connectivity changes related to tau pathology compared to the NODDI model.
• Validated the method on datasets from Human Connectome Project (HCP) subjects and Autosomal Dominant Alzheimer's Disease (ADAD) patients.

Conclusions:

• The novel diffusion signal decomposition method provides a more effective way to study cortical microstructure and connectivity.
• This approach offers improved insights into neurodegenerative diseases like Alzheimer's by detecting subtle tau-related pathology.
• The publicly available codebase facilitates further research in neuroimaging analysis.