Abstracts

Rationale: Focal drug-resistant epilepsy may require intracranial monitoring for pre-surgical evaluation. Resection has variable rates of seizure freedom (Tellez-Zenteno et al., Epilepsy Res 2010;89:310-318). Electrocorticography (ECoG) data are analyzed by visual inspection but there is often a lack of quantitative analysis. Tools to improve identification of the EZ may be useful to guide resection and improve outcomes. EZTrack is an analytic system based on network data science and dynamical systems models designed to provide quantitative analysis of ECoG data. EZTrack aims to identify the EZ by determining the fragility of each electrode. Fragility of an electrode is defined as the smallest change or perturbation in its functional connections to its neighbors that is needed to cause an unstable network thus leading to seizure. Our hypothesis is that EZTrack analysis of ECoG will determine the most fragile electrodes in the epileptic network that corresponds to the EZ. Methods: We retrospectively analyzed ECoG data from 6 patients at the University of Maryland Medical Center in a pilot study using EZTrack. Three typical seizures were chosen for analysis as most patients had at least 3 seizures during monitoring. Analysis of each seizure began 60 seconds before onset and ended 60 seconds after seizure cessation. ECoG data was collected with subdural grids/strips and analyzed by a board-certified epileptologist. Data was reconstructed into a sequence of stable dynamical network models and analyzed via EZTrack to produce fragility values ranging from 0 to 1 with 0 correlating with no fragility and greatest fragility with a value of 1. Fragility of each electrode was graphed with respect to time on a heat map. Red corresponds to the greatest areas of fragility and blue corresponds to no fragility. A row sum of fragility was calculated for each electrode to show which electrodes appeared most fragile. The most fragile electrodes were compared to the outline of the EZ as determined in routine clinical assessment. Results: Time since ECoG monitoring for all 6 patients is as follows: range: 2.3-7 years, mean 4.65 years; median 4.45 years. Five patients underwent resection. One patient did not have resection due to overlap of EZ with functional cortex. Five patients are now seizure free (Table 1). EZTrack correlated with the clinical read of the ECoG of 4 patients and only partially correlated with 2 patients by visual inspection. Of the 2 patients that only partially correlated (Patients 3 and 5), EZTrack identified areas of fragility that were not felt to represent epileptic activity through clinical assessment. Conclusions: Accurate localization of the EZ is important to improve clinical outcomes after intracranial monitoring for surgical resection. Quantitative analysis may help to further define the EZ prior to surgical resection. EZTrack can identify the most fragile electrodes within an EZ. Funding: Maryland TEDCO 0715-014_3