Abstracts

Rationale: InterIctal Spikes (IISs) are known as biomarkers used in identifying seizure onset zones in focal epilepsy. Therefore, it is essential to develop algorithms that can localize these events and classify them based on their types. These pathological discharges have been investigated extensively in EEG signals; however, they have been scarcely investigated in intracranial recordings (iEEG) that could be associated with the high variability of these spikes in their morphologies in intracranial recordings. Hence, the quantitative analysis of their morphology may assist in surgical planning and improve surgical outcomes. This study investigates the relationship between the prominence of InterIctal spikes and their spatial distribution relative to the lobes and hemispheres. We hypothesize that IISs can be discriminated in anterior and posterior and mesial versus neocortical temporal spikes using their Spectro-temporal characteristics.

Methods: We analyzed the intracranial recordings obtained from patients with medically resistant epilepsy (MRE) implanted with Depth Electrodes (DE) at the Epilepsy Monitoring Unit of the Epilepsy Program at Western University. The data were cleaned, denoised, montaged and analyzed based on the clinical annotations' segment, removing sleep intervals, and observed Ictals. Pathological spikes are detected after filtering in frequency bands (20-70) Hz. Then, the signal waveform and its power were extracted symmetrically in non-overlapping intervals of 500 ms. Next, the statistical properties of these spikes and the best fit distributions were calculated for hippocampus electrodes. Finally, the algorithm discriminates these spikes based on their energy waveform distributions based on the electrode's locations. Our algorithm calculates spike waveform energy information in anterior vs posterior and left vs right using AIC and Loglikelihood criteria.

Results: Data included eight sessions of 24 hours of extracellular recordings from two patients with MRE with a history of bitemporal lobe epilepsy. More than 210 hours were extracted from four hippocampus electrodes anterior, posterior and left/right electrodes. Our results indicate that detected spikes on the hippocampus contacts in the left hemisphere are significantly different in anterior than posterior and in each hemisphere. We performed the parametric Kolmogorov-Smirnov test to assess statistical significance, and a value of P < 0.05 was considered statistically significant. The algorithm uses spatiotemporal characteristics of detected spikes before and after each, classifies them based on their energy waveform’s information, and differentiates them based on AIC and Loglikelihood estimators.