Multiscale modes of functional brain connectivity

Area of Science:

• Neuroscience
• Computational Biology
• Medical Imaging

Background:

• Brain information processing involves localized and distributed systems operating at multiple scales.
• Current functional brain connectivity methods often miss cross-scale interactions.
• Existing approaches use limited modes or parcellations, failing to capture multiscale dynamics.

Purpose of the Study:

• To introduce Multiscale Probabilistic Functional Modes (mPFMs) for comprehensive functional brain connectivity analysis.
• To enable direct estimation of functional connectivity within and across different scales of brain organization.
• To develop enhanced functional MRI (fMRI) biomarkers for personalized traits and diseases.

Main Methods:

• Data-driven multilevel Bayesian modeling applied to large functional MRI (fMRI) population and individual data.
• Development of a novel mapping (mPFMs) comprising modes at various scales of granularity.
• Validation using simulations and real UK Biobank data.

Main Results:

• mPFMs emerged from data, capturing both distributed brain modes and their subcomponents.
• The new mapping enables direct estimation of within- and across-scale functional connectivity.
• mPFMs achieved higher accuracy in predicting ~900 personalized traits from UK Biobank data compared to standard techniques.

Conclusions:

• mPFMs provide a new framework for functional connectivity modeling, integrating information across multiple brain scales.
• This approach offers a more complete understanding of brain function and its relation to individual traits.
• mPFMs can yield enhanced fMRI biomarkers for predicting traits and potentially identifying diseases.