A computational tool for pre-surgical evaluation of epilepsy patients

Abstract

Epilepsy is a neurological condition affecting 50 million people worldwide. Patients suffer from recurrent seizures that disrupt various aspects of normal life and lead to brain damage and other comorbidities. Unfortunately, up to 40% of epilepsy patients have medically refractory epilepsy (MRE), which is unresponsive to medication. For these patients, the only solution is surgical removal of the region of the brain causing the seizures, or the epileptogenic zone (EZ). However, surgical resection is contingent upon correct identification of the EZ, a challenging task that is currently done by visual inspection of hundreds of channels of information. Unsurprisingly, resection has a 50% long-term failure rate, largely due to misidentification of the EZ. Repeated invasive monitoring to increase the accuracy of EZ identification is also not a desirable option because it is both dangerous and expensive. We have developed a computational tool, EZTrack, to help clinicians identify a patient's EZ using intracranial EEG data. EZTrack employs a novel algorithm to analyze the brain as a network during different seizure and non-seizure time periods in order to isolate the likely location of the EZ. In preliminary testing of EZTrack on data from 19 patients who underwent surgical resection, we were able to correctly predict all of their outcomes. EZTrack holds far-reaching implications for epilepsy patients, including reduced hospital stays, decreased costs, lowered risk of comorbidities, and better surgical outcomes. With EZTrack, patients who undergo the formidable procedure of surgical resection will have a significantly greater chance of actually living life seizure-free.