HFA separated SOZ pairs in detected spike-free HUP SEEG windows (median AUC 0.65)

Interictal spikes contain 80–150 Hz power, so high-frequency activity (HFA) can act as a spike proxy when it separates seizure-onset-zone (SOZ) from non-SOZ pairs. We tested whether this separation remained in 1-s windows without detected spikes in 120-s interictal SEEG clips from 32 HUP patients in OpenNeuro ds004100 v1.1.3. With within-shaft bipolar pairs, clinical SOZ labels, and a pre-analysis plan, we used a 10–55 Hz amplitude-threshold detector (6 robust SD; stricter check: 4 SD) and measured each subject’s AUC for log 80–150 Hz power. Local exclusion of windows with a spike in the pair or adjacent pairs gave a median AUC of 0.65, above 0.5 in 27 of 32 subjects (Holm-adjusted Wilcoxon p = 0.0004). Global exclusion was possible in 18 subjects and gave a median AUC of 0.68 (Holm p = 0.047). Removal of spike windows changed the median per-subject HFA AUC by less than 0.01; with the stricter detector, the local AUC was 0.66. These results indicate that HFA separation in these clips was not only an effect of detected spikes. The detector was not validated against expert spike marks, small spikes could remain, and 0.25-s padding cannot remove longer after-effects. The clips were short, sleep/wake state was unknown, and all subjects came from one center. Global exclusion could bias results because it was possible for only 18 of 32 subjects. HFA includes all 80–150 Hz power, not only discrete high-frequency oscillations.

interictal SEEG · high-frequency activity · seizure-onset zone · spike detection · HUP · OpenNeuro ds004100 · AUC · epilepsy