Modeling Neural Collaboration: Measuring the Odds of Cooperation in the Gyri of Frontal Lobes

The analyses of brain connectivity networks with small-world metrics motivates the use of exponential random graph models (ERGMs) for generating insight into the complex neurobiological interactions for many neurological conditions and disorders.  A .NET/R based research tool for neuroinformatics is presented that provides an exploration of which network metrics best characterize brain networks in the Inferior Frontal Gyrus, Middle Frontal Gyrus, and Superior Frontal Gyrus regions of the frontal lobes.  Explanatory network metrics such as connectedness, local clustering/Efficiency, Global Efficiency, Degree Distribution and Location are used as predictors for testing significance of these predictors across multiple subjects.  ERGMs with the above predictors are estimated with MCMC MLE for each subject and gyri. Model selection and goodness of fit strategies are examined for robustness across subjects.  Potential clinical applications exist for ERGMs because they provide the ability to efficiently represent complex brain networks by examining the relationship between global and local brain structure.  The results here demonstrate that not only can this tool be effective in modeling brain networks, but can have clinical applications as well.