Brain Network Structure and Interventions in a Computational Model of Epilepsy

Some forms of drug-resistant epilepsy can only be treated via surgical intervention. This form of treatment requires the removal of a part of the brain identified as the seizure source. Current methods for surgical treatment are risky and many times unsuccessful. A deeper understanding of how brain connectivity facilitates seizure propagation is necessary for developing improved surgical techniques. Experimental limitations make certain clinical investigations of epilepsy difficult or impossible, but computational modeling offers a way forward when experimentation in living systems is impractical or unsafe. We used a full-hemisphere computational model for epilepsy to investigate the role of network structure in facilitating seizure propagation. From this model, we derived a novel network measure that was used to predict nodes with high epileptic influence. This measure was shown to outperform other common network measures that are widely used to characterize spreading and seizures in networks. Further investigation showed that this measure can be used to inform simulated interventions for seizure suppression. Our results suggest that this measure could be used in combination with individualized connectivity data from epileptic patients to inform possible routes for surgical intervention