Resolving structural variability in network models and the brain.

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Human cognitive function relies on complex white matter pathways.
• Mechanistic drivers of brain network architecture remain poorly understood.
• Identifying generative principles for brain connectivity is challenging.

Purpose of the Study:

• To contrast empirical human brain network properties with synthetic network models.
• To investigate the roles of physical embedding and temporal growth in network formation.
• To identify generative mechanisms underlying brain network architecture.

Main Methods:

• Utilized diffusion spectrum imaging data from the human brain.
• Developed and analyzed 13 synthetic network models with varying growth rules and constraints.
• Characterized empirical and synthetic networks using graph metrics (degree distribution, assortativity, hierarchy, scaling properties).

Main Results:

• Synthetic models constrained by physical embedding in anatomical regions most closely matched human brain network properties.
• Network models designed for specific properties (e.g., assortativity) did not simultaneously exhibit others (e.g., hierarchy).
• Network properties appear relatively independent, suggesting diverse underlying neurobiological mechanisms.

Conclusions:

• Physical embedding is a key factor in generating realistic brain network structures.
• Multiple neurobiological mechanisms likely contribute to the development of brain network architecture.
• Developed network models offer a framework for statistical inference of brain network structure from neuroimaging data.