Group-common and individual-specific effects of structure-function coupling in human brain networks with graph neural...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Brain Imaging

Background:

• The human cerebral cortex exhibits functional segregation and synchronization, linked by white matter structural connectivity.
• Understanding the balance between group-common and individual-specific structure-function coupling is crucial for cognitive development and neuropsychiatric disorders.

Purpose of the Study:

• To investigate the extent to which brain structure predicts brain function across individuals.
• To determine if structure-function coupling is a group-common characteristic or varies individually.
• To explore the hierarchical organization of coupling effects across the cortex.

Main Methods:

• Utilized two independent, high-quality neuroimaging datasets.
• Employed graph neural networks (GNNs) to predict functional connectivity from structural connectivity.
• Compared GNN performance against traditional correlation approaches.

Main Results:

• Graph neural networks accurately predicted individual functional connectivity from structural connectivity, demonstrating strong coupling.
• Structure-function coupling was primarily driven by network topology and outperformed correlation methods.
• Coupling effects were predominantly group-common, with significant, albeit subtle, individual-specific components.
• Regional coupling effects showed hierarchical organization along a sensorimotor-association axis, with higher individual effects in association cortices.

Conclusions:

• Brain structure-function coupling is a robust phenomenon, largely shared across individuals.
• Individual differences in structure-function coupling are subtle but significant and hierarchically organized.
• Considering both group and individual coupling effects is vital for understanding brain organization and individual variability.