In 35 HUP patients, spike-leader fraction did not mark the clinical SOZ beyond spike rate

We tested whether the first bipolar pair to cross a spike threshold in a multi-pair event marks the clinical seizure onset zone (SOZ) in interictal SEEG. We analyzed run-01 segments from 37 HUP patients identified in OpenNeuro ds004100; 35 met the inclusion criteria, with 76–240 s per patient. Adjacent bipolar pairs were formed from good channels. A 10–60 Hz amplitude detector with a 6 × MAD threshold and 200 ms refractory period identified spikes; events clustered detections within 50 ms, and the earliest threshold crossing was the leader. Patient-level AUCs compared SOZ and non-SOZ pairs, with sign-flip tests and Holm correction. Leader rate separated SOZ pairs from non-SOZ pairs (mean AUC 0.704; 31/35 above 0.5; Holm p = 1.5 × 10⁻⁵), but was lower than plain spike rate (AUC 0.740; mean difference −0.036; Holm p = 0.0047). Leader fraction (leads divided by participations) was at chance (mean AUC 0.500, n = 29, p = 0.995). Results were similar for 30 ms and 100 ms event windows and in the Engel I subgroup (n = 17). In these short interictal segments and under clinical SOZ labels, SOZ pairs led more events because they had more spikes; threshold-crossing order added no SOZ information. The amplitude detector lacked a morphology check, threshold-crossing time is not a true propagation onset, and events could include pairs on different shafts.

stereo-EEG · epilepsy · seizure onset zone · spike detection · interictal · AUC · OpenNeuro