Abstract

For approximately 20 million people with drug-resistant epilepsy, recurring, spontaneous seizures have a devastating impact on daily life. The efficacy of surgical treatment for controlling seizures is hindered by a poor understanding of how some seizures spread to and synchronize surrounding tissue while others remain focal. To pinpoint network regions that regulate seizure evolution, we present a novel method to assess changes in synchronizability in response to virtually lesioning cortical areas in a validated computational network model. In human patients implanted with electrocorticographic sensors, we apply our virtual cortical resection technique to time-varying functional networks and identify control regions that synchronize or desynchronize cortical areas using an antagonistic push-pull control scheme to raise or lower synchronizability. Our results suggest that synchronizability before seizures predicts seizure evolution: in focal seizures, the strongest controllers are located outside seizure-generating areas. These methods,while applied here to epilepsy, are generalizable to other brain networks, and have wide applicability in isolating and mapping functional drivers of brain dynamics in health and disease.