Prediction of multifocal epileptogenic zones using normalized transfer entropy

Stereo-EEG with multiple intra-cerebral depth electrodes is a pre-surgical tool for monitoring and analyzing the electrophysiological signals from the brain. The standard method of evaluating these signals by visually identifying the changes in the EEG signal is being slowly replaced by the transfer entropy that accurately accounts for non-linearity and incorporates dynamic interactions between the systems. Partial epilepsy can arise from a single focus or have multifocal onset. Knowing the number of epileptic foci and their location is essential to guide the choice of treatments recommended to the patient. Epilepsy that initially appears to be unifocal, based on simple visual analysis, may turn out to be multifocal on more detailed evaluation. In this paper, we computed the directional information transfer for all possible pairs of the 80 signals that were collected from the brain of an epileptic patient. We have localized the epileptogenic zones (EZs) and identified multiple seizure focal points in the patient from the plots obtained. Understanding such multifocal onset patients and avoiding focused surgical resections in such patients may be important to avoid surgical failures and morbidity.