Network Analyses in Epilepsy

Focal epilepsy is now perceived as a multiscale network disorder. Using network-based graph theoretical approaches, investigators have begun to investigate the widespread functional and structural effects of epilepsy and seizures on the brain.^1,2 Graph theory defines networks as sets of nodes, or brain regions, and edges, the connections between nodes. In this mathematical framework, the topologic properties of nodes and edges can then be studied at varying scales, ranging from nodal to modular to global levels, and visualized as connectivity matrices.³ The figure demonstrates 2 example analysis pipelines using a graph theoretical framework: diffusion tractography (figure, A) and intracranial EEG (figure, B). Multiple recent studies have used these analyses to predict epilepsy surgery outcomes using presurgical networks,⁴ with many leveraging the ability of machine learning algorithms to investigate a large number of edges per patient.⁵