Multi-hierarchy Network Configuration Can Predict Brain States and Performance.

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• The brain exhibits a hierarchical modular organization that shifts with functional states.
• Understanding multi-hierarchy network configuration is crucial for deciphering brain function.
• Existing methods struggle to capture the complexity of multi-level brain network organization.

Purpose of the Study:

• To introduce an eigenmodal decomposition approach for detecting multi-hierarchy modules in brain networks.
• To quantify network configuration changes across hierarchical layers using novel metrics.
• To investigate the differences in brain network configuration between resting and task states.

Main Methods:

• Developed an eigenmodal decomposition method to identify hierarchical brain network modules.
• Defined and utilized metrics: node configuration matrix, separability, and combinability.
• Applied the method to simulated random networks and human fMRI data (resting and task states).

Main Results:

• The proposed method successfully identified hierarchical submodules consistent with brain structure.
• Real brain networks showed higher separability and lower combinability than random networks.
• Task states exhibited less separability and greater combinability between modules compared to the resting state.
• Brain network configuration, especially during tasks, predicted cognitive performance and individual differences.

Conclusions:

• The eigenmodal decomposition approach offers novel insights into complex brain network organization at multiple hierarchies.
• Task-induced brain network attributes are more powerful in characterizing and predicting behavioral traits than resting states.
• This study advances our understanding of brain mechanisms and individual differences through multi-level network analysis.