A Stepwise Multivariate Granger Causality Method for Constructing Hierarchical Directed Brain Functional Network.

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Brain functional networks are complex and hierarchical, facilitating segregated functions and global integration.
• Directed network construction offers causal insights into brain region relationships.
• Existing methods may not fully capture the hierarchical architecture of directed brain networks.

Purpose of the Study:

• To propose a novel approach, stepwise multivariate Granger causality (SMGC), for modeling directed and hierarchical brain functional networks.
• To explore stepwise causal relationships within complex brain networks.
• To investigate the hierarchical organization of brain networks using functional magnetic resonance imaging (fMRI) data.

Main Methods:

• Developed the stepwise multivariate Granger causality (SMGC) method.
• Conducted simulation studies to validate SMGC's ability to capture hierarchical structures.
• Applied SMGC to resting-state fMRI datasets to analyze brain network architecture.

Main Results:

• SMGC successfully captured multiple levels of hierarchy in directed brain networks.
• Analysis of fMRI data revealed within-network directed connections and between-network pathways in hierarchical networks.
• The default mode network (DMN) plays a key role as a causal source and relay station.
• Athletes showed enhanced bidirectional communication between DMN and central executive network (CEN), and directed connections from salience network (SN) to CEN.

Conclusions:

• The SMGC approach effectively models the hierarchical architecture of directed brain functional networks.
• This method provides new insights into stepwise causal relationships within the brain.
• SMGC has potential applications in understanding altered brain network hierarchies in neuropsychiatric disorders.