Mapping Functional Brain Organization Using Artificial Intelligence.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Imaging

Background:

• The human brain's distinct regions are defined by cytoarchitecture, structure, function, and connectivity.
• Brain parcellation into homogeneous regions is vital for understanding brain organization and function.
• Functional parcellation using fMRI data reveals brain network changes across the lifespan and in disease.

Purpose of the Study:

• To review current AI-driven methodologies for functional brain parcellation using rs-fMRI.
• To compare traditional and AI-based approaches in functional parcellation.
• To discuss validation strategies, limitations, and future directions for AI in brain mapping.

Main Methods:

• Review of supervised, unsupervised, and self-supervised learning frameworks in functional parcellation.
• Comparison of traditional methods (e.g., ICA) with AI methods (e.g., GNNs, CNNs, Transformers).
• Elaboration on validation strategies: reproducibility, functional homogeneity, task-based fMRI/electrophysiology alignment, and cross-modality validation.

Main Results:

• AI enables data-driven and individualized functional brain mapping.
• AI methods offer advanced spatial and temporal feature extraction for brain parcellation.
• Various AI techniques show promise in mapping brain organization.

Conclusions:

• AI significantly transforms functional parcellation, enabling more precise and individualized brain mapping.
• Future directions include multimodal integration, foundation models, and explainable AI for enhanced brain understanding.
• AI-powered functional parcellation supports research into brain organization across the lifespan and in disease states.