A flexible graphical model for multi-modal parcellation of the cortex.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Mapping

Background:

• Neuroimaging advances provide extensive in-vivo brain organization data.
• Current brain parcellation methods often rely on single imaging modalities, limiting comprehensive analysis.
• Integrating multiple modalities is crucial for accurate and robust cortical subdivisions.

Purpose of the Study:

• To introduce Graph-based Multi-modal Parcellation (GraMPa), an iterative framework for multi-modal brain parcellation.
• To develop a method that effectively fuses information from diverse neuroimaging modalities.
• To improve the accuracy and robustness of brain cortical organization mapping.

Main Methods:

• Developed an iterative framework (GraMPa) to handle multiple input modalities.
• Fused parcellations from different modalities based on local reliability at each iteration.
• Explored group-wise parcellation using resting-state fMRI, diffusion MRI, myelin maps, and task fMRI.

Main Results:

• The GraMPa framework demonstrated convergence towards mutually informed, modality-specific parcellations.
• Multi-modal integration showed stronger agreement with established brain atlases.
• Resulting connectivity networks were more robust and representative of the population.

Conclusions:

• Integrating multiple neuroimaging modalities significantly enhances brain parcellation accuracy and robustness.
• GraMPa provides a flexible framework for leveraging diverse data in brain mapping.
• The approach leads to improved population-based representations of brain connectivity.