Predicting Surgical Outcomes in Epilepsy Patients Using Directed Transfer Function and Computational Model

Abstract

For patients with medically refractory epilepsy, surgical resection of the epileptogenic zone is one of the effective treatments. The commonly used method is based on the clinician's experience to localize the epileptogenic zone, but there are still some patients without achieving seizure-free after surgery. Therefore, predicting the outcome of surgical treatment may play a key role in subsequent treatment. Epileptic networks using dynamic computational models were used to simulate the seizure process of epilepsy, which could be used to predict the surgical outcome. In this paper, we investigate whether a computational network with causal correlation, instead of undirected correlation, can improve the accuracy of prediction. The directed transfer function (DTF) was used to construct the causal network based on the interictal electrocorticogram (ECoG) from five patients. The outcomes of three patients were predicted correctly, including one who had failed to predict by using the undirected network. This preliminary result suggests that our proposed method using DTF and computational modelling may further improve the accuracy of outcome prediction.