Two interictal clips reproduce amplitude-feature SOZ rankings, while spike-rate rankings improve when averaged

We tested whether seizure onset zone (SOZ) feature values from one 30 s interictal SEEG clip reproduce in a second clip. In 36 HUP patients, 3,687 matched within-shaft bipolar pairs included 392 SOZ pairs. Broadband power, high-gamma power, and line length had median within-patient clip correlations of 0.96–0.97 and per-patient AUC intraclass correlations of 0.87–0.94. Their mean SOZ AUCs were 0.65–0.68 in each clip, and averaging clips did not improve AUC. Spike-like transient rate was less stable (median correlation 0.35; AUC ICC 0.37); its mean AUC increased from 0.66 to 0.71 when the two clips were averaged (change +0.052, 95% bootstrap CI +0.029 to +0.078; Holm-adjusted p = 0.002). The 30–45 Hz aperiodic exponent did not identify SOZ channels (AUC 0.48 and 0.47), although its per-patient AUCs agreed between clips (ICC 0.77; same side of 0.5 in 78% of patients). These findings indicate that amplitude-feature rankings are stable across the two clips, while event-rate estimates benefit from averaging. The study used only the first 30 s of each clip at one center; time between clips and vigilance state were unknown, clinical SOZ labels were the reference, tissue labels were unavailable, and the simple spike detector was not validated.

stereo-EEG · seizure onset zone · interictal · test-retest reliability · AUC · feature stability · epilepsy