Computational modeling of whole-brain dynamics: a review of neurosurgical applications

Abstract

A major goal of modern neurosurgery is the personalization of treatment to optimize or predict individual outcomes. One strategy in this regard has been to create whole-brain models of individual patients. Whole-brain modeling is a subfield of computational neuroscience that focuses on simulations of large-scale neural activity patterns across distributed brain networks. Recent advances allow for the personalization of these models by incorporating distinct connectivity architecture obtained from noninvasive neuroimaging of individual patients. Local dynamics of each brain region are simulated with neural mass models and subsequently coupled together, considering the subject’s empirical structural connectome. The parameters of the model can be optimized by comparing model-generated and empirical data. The resulting personalized whole-brain models have translational potential in neurosurgery, allowing investigators to simulate the effects of virtual therapies (such as resections or brain stimulations), assess the effect of brain pathology on network dynamics, or discern epileptic networks and predict seizure propagation in silico. The information gained from these simulations can be used as clinical decision support, guiding patient-specific treatment plans. Here the authors provide an overview of the rapidly advancing field of whole-brain modeling and review the literature on neurosurgical applications of this technology.