Clustering Epileptic Spike Waveforms for Predicting Seizure Onset Zone

Interictal spikes, observed through electrocorticography (ECoG), offer critical insights that assist in the surgical planning for patients with drug-resistant epilepsy. However, the temporal dynamics and shape of these spikes remain unclear. To address this issue in support of epilepsy surgery, our study aimed to analyze the shape of interictal spikes and measure their distance to the seizure onset zone (SOZ). Twelve patients' data from the iEEG portal were retrospectively studied. For analysis, half an hour of ECoG data was used from each patient, with the data being truncated before the onset of a seizure. Spikes were first detected and grouped in a sequence, then clustered into interictal epileptiform discharges (IEDs) and non-IED groups using a clustering algorithm. The distance of the spikes from IED and non-IED groups to SOZ was quantified and compared using the Wilcoxon rank-sum test. Spikes in the IED group were often in SOZ or close to it, while spikes in the non-IED group were within the distance of the SOZ or non-SOZ area. At the group level, the distribution for sharp wave, positive baseline shift, slow wave, and slow wave to sharp wave ratio was significantly different for IED and non-IED groups. The distance of the IED cluster was 23.36mm and significantly closer to the SOZ than the 37.52mm for non-IEDs. These findings offer valuable insights into the shape and spatiotemporal dynamics of spikes, which could help localize the SOZ and assist neurosurgeons in making more informed decisions for epilepsy surgery.