Node-reconfiguring multilayer networks of human brain function.

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Functional brain network analysis relies on defining network nodes, often using static Regions of Interest (ROIs).
• Static ROIs, collections of functional magnetic resonance imaging (fMRI) voxels, fail to capture dynamic brain activity and temporal reconfigurations.
• This static definition leads to low functional homogeneity and violates assumptions in network analysis, potentially causing spurious network structures.

Purpose of the Study:

• To introduce a novel node-reconfiguring multilayer network model for brain functional networks.
• To optimize ROI boundaries dynamically for high functional homogeneity within each time window.
• To investigate the impact of dynamic ROI reconfigurations on brain network modeling and functional homogeneity.

Main Methods:

• Developed a node-reconfiguring multilayer network model where network layers represent time windows.
• Optimized ROI boundaries in each time window for maximal functional homogeneity.
• Quantified intralayer functional connectivity and interlayer ROI overlap across time windows.

Main Results:

• The ROI optimization approach achieved over a 10-fold increase in highly homogeneous ROIs compared to static atlases.
• Optimized ROIs exhibited non-trivial reorganization across short and extended time scales.
• Intermediate levels of ROI reorganization correlated with stronger intralayer functional connectivity (hubness).

Conclusions:

• Dynamically reconfiguring parcellations provide more accurate network models of brain function.
• This approach supports the chronnectome paradigm, viewing the brain as continuously reconfiguring sources.
• Optimized ROIs enhance functional homogeneity, addressing limitations of static definitions in fMRI network analysis.