Abstracts

Rationale: Several epileptic biomarkers have been proposed to predict the seizure onset zone (SOZ). Among these, spikes, high-frequency oscillations (HFOs), sequences of spikes, and combinations of spikes with HFOs have shown promise in the literature for SOZ prediction. So far, the rate of these events per minute has been utilized to predict the SOZ. In this study, we aim to demonstrate that the spatial distance of each individual event can serve as an alternative method for identifying the SOZ.


Methods: This research investigated two patients diagnosed with focal epilepsy who underwent extra-operative resting-state electrocorticography (ECoG) recording at Megumino Hospital in Japan. The data were collected when subjects were in their resting state of sleep at night. The SOZ was identified based on ECoG data and validated using video EEG recordings. Five biomarkers, spike, ripple HFO, fast ripple HFO, pathological HFO (pHFO), and spike sequence, were used for the localization of SOZ. An automated event detection was employed to identify spikes. Sequential spikes occurring closely in time and across multiple electrodes were identified as spike sequences, with the initial spike designated as the sequence leader. Spikes occurring within 50 milliseconds of the leader or within 15 milliseconds of a preceding spike were incorporated into the sequence. Sequences with fewer than five spikes were discarded. Later, an HFO detection algorithm was used to detect ripple and fast ripple events. Subsequently, the temporal occurrence of spikes and ripple HFOs were compared to identify any coinciding events called pHFO. For distance calculation, channels containing at least one event were selected, and the minimum distance from each selected channel to channels indicating the seizure onset area was computed using the Euclidean distance measure.


Results: All biomarkers were automatically detected across 280 electrodes in three patients, with a total data duration of 68 minutes. The detection yielded 1334 spikes, 370 ripples, 19 fast ripples, 175 pathological HFO, and 783 spike sequence events per minute. By computing the Euclidean distance between these events and the SOZ, average distances of 32.7±29.1, 41.7±27.5, 61.5±23.1, 27.4±23.8, and 28.7±25.2 per millimeter were calculated for spikes, ripples, fast ripples, pHFO, and spike sequences, respectively. For the best biomarkers of pHFO and spike sequence, 25.52% and 26.07% of events were inside the SOZ area.


Conclusions: Automatic detection and localization of interictal epilepsy biomarkers can serve as an alternative method for identifying the SOZ. Thus, achieving a good overlap with the SOZ is important. This study revealed that pathological HFOs and spikes within sequences may be particularly effective biomarkers for localizing seizure onset due to their proximity to the SOZ. From a clinical perspective, precise localization is important, and the 25.52% overlap provides a very good starting point for subsequent events review to pinpoint the epileptogenic region. Notably, it is important to understand the clinical impact of pHFO events and spike sequences outside the SOZ and how they should be treated in the future.


Funding: no funding