Abstracts

Rationale: Identification of the seizure onset zone (SOZ) using intracranial electrocorticography (ECoG) is a complex task relying on pattern recognition by an expert interpreter. Various ictal and interictal patterns can signal a seizure focus. Disagreement exists as to which patterns best correlate with successful surgical outcomes. High-current single pulse electrical stimulation (SPES) that is manually delivered to ECoG electrodes has been shown to evoke afterdischarges preferentially within the SOZ. High stimulus intensities, however, can cause seizures and patient discomfort. The manual delivery of the stimulation and the variability of the responses have prevented clinical adoption of this technique. We show here that automated delivery of low-current SPES to epileptogenic tissue evokes distinct patterns of neuronal activity that can be used for rapid identification of the SOZ. We use a unique feature, delayed high-frequency spectral suppression, to provide an estimate of the SOZ. Methods: We recorded responses to automated low-current SPES from five patients with intractable epilepsy who were undergoing long-term ECoG monitoring for SOZ determination. Stimulation amplitude was set at 3.5 mA using monopolar biphasic pulses of 1 ms duration. Pulses were delivered at an average interval of 3 sec using a pseudorandom sequence to increase the spatiotemporal distribution of charge and avoid potentiation. At least 10 trials were applied to a total of 339 implanted ECoG electrodes for evaluation of trial-averaged responses on nearby electrodes. Delayed high-frequency suppression was measured by calculating the z-score normalized trial-averaged time-frequency power from 0.4 to 1 sec and 70 to 250 Hz post-stimulation. Normalization and bootstrap comparisons for the calculation of significance were performed using the 1 sec of data immediately prior to each stimulation pulse. Electrodes that recorded suppression that was greater than half the maximum suppression were considered the estimated SOZ. The estimated SOZ was then compared with the clinical SOZ using the Fisher Exact test to determine overall significance (p Results: The estimated SOZ, based on the analysis of delayed high-frequency suppression, significantly identified the clinical SOZ in all five patients. Stimulation of electrodes inside the SOZ typically showed broadband or high-frequency suppression of power lasting as long as 1 sec on nearby electrodes. Suppression was also observed outside the SOZ in some patients, but these responses were typically shorter in duration. Responses showing increases in power generally occurred when stimulation was applied outside the SOZ. Importantly, data collection and analysis required less than 2 hours of patient time, the stimulation was undetectable by patients, and no seizures were induced. Conclusions: Using a unique response feature, we show that automated low-current SPES can be used to provide estimates of the clinical SOZ in a small patient cohort. We believe this approach could increase the safety, speed, and reproducibility of SOZ identification while reducing cost, subjectivity, and patient discomfort. Funding: none