In 36 SEEG patients, the 40–150 Hz exponent separated clinically marked SOZ channels; the 30–45 Hz exponent did not

This analysis tested whether interictal aperiodic exponents identify clinically marked seizure-onset-zone (SOZ) channels in stereo-EEG (SEEG). It used 37 patients and 3,789 bipolar channels per run from OpenNeuro ds004100 (HUP), analyzing the first 60 s of two 300-s interictal clips. In the primary run, the 40–150 Hz exponent had a median per-patient SOZ-versus-non-SOZ AUC of 0.725 (31 of 36 patients above 0.5; Holm-adjusted p=0.0003); the independent second run replicated the effect (median AUC 0.715, p=0.0003). The effect remained within electrode shafts (median AUC 0.825, 29/31 patients, p=0.0003). The 30–45 Hz exponent did not separate SOZ channels (median AUC 0.471, p=0.25). Low-frequency power and line length were also higher, while relative high-gamma power was lower, in SOZ channels. The 40–150 Hz exponent correlated with 1–40 Hz power (median rho 0.81); after a post hoc power control, its effect was smaller but remained (median AUC 0.58, p=0.034 in run 1; 0.61, p=0.032 in run 2). No feature separated resected from non-resected channels after correction in seizure-free patients, and SOZ AUC did not differ between seizure-free and not-seizure-free patients. Limitations include short clips, clinical SOZ labels, no anatomical control beyond within-shaft analysis, the simple log-log spectral fit, no correction across analysis families, and a post hoc power control.

stereo-EEG · seizure-onset zone · aperiodic exponent · spectral power · interictal · epilepsy