Early-ictal SEEG pair maps show moderate seizure-to-seizure agreement in 35 patients

We tested whether the same bipolar SEEG pairs show the largest early-ictal 30–100 Hz power rise across seizures in each patient, without using clinical seizure-onset-zone labels. The analysis used archived per-seizure features from 35 HUP patients with 2–3 seizures each (99 seizure pairs). The mean between-seizure Spearman correlation was 0.52 (mean Fisher z 0.571, 95% CI 0.41–0.74); 31 of 35 patients had positive agreement (p=5×10⁻⁵). Agreement remained after removing the rank of pre-ictal 30–100 Hz power (partial mean z 0.558; Holm-adjusted p=0.0002). The mean top-five overlap was 41%, compared with 6% chance, although 27 of 99 seizure pairs shared no top-five pair. Agreement did not differ between seizure-free and failed patients (mean z 0.59 vs 0.55; Holm-adjusted p=1.0) and did not correlate with per-patient SOZ AUC (rho=0.11, n=34; Holm-adjusted p=1.0). Pre-ictal power itself was highly consistent (rho=0.96). These findings show moderate repeatability and support multi-seizure averaging, while indicating that a single-seizure map may vary substantially. There was no non-seizure control window; repeating arousal or artifact at onset remains possible. The study included at most three seizures per patient, one center, no artifact rejection, and had low power for outcome tests. The exact OpenNeuro dataset release used for the archived feature extraction was not recorded.

stereo-electroencephalography · epilepsy · seizure onset · high-frequency power · reproducibility · intracranial EEG · seizure outcome