Interictal SEEG aperiodic exponent did not identify seizure-onset channels

In 36 HUP patients with SEEG (3,640 bipolar channels; 35 patients had both SOZ and non-SOZ channels), the primary 30–45 Hz interictal aperiodic exponent did not distinguish clinical seizure-onset-zone (SOZ) channels (mean within-patient AUC 0.476, bootstrap 95% CI 0.419–0.532; 17/35 AUCs >0.5; sign-flip p=0.4181). A 2–45 Hz fit also showed no effect (AUC 0.548, CI 0.478–0.616; p=0.1922). Log broadband and high-gamma power and relative delta power were higher in SOZ channels; relative alpha and beta power were lower (Bonferroni-corrected p≤0.0258). Theta, gamma, and relative high-gamma power were not significant after correction. No feature differed significantly between seizure-free (n=20) and not seizure-free (n=15) patients (all uncorrected permutation p≥0.069). In seizure-free patients, resected-channel relative delta and beta AUCs were 0.670 (p=0.0002) and 0.365 (p=0.0054); the 30–45 Hz exponent AUC was 0.419 (p=0.0369, uncorrected), opposite to a steeper-SOZ hypothesis. These results concern 15 seconds of interictal run-01 data per patient, median Welch PSD, and specified linear fits. Short recordings make PSD estimates noisy. Anatomical differences and unremoved interictal discharges can explain some power effects; no anatomical normalization was applied. The results are not a clinical localization method. Other fit methods, frequency ranges, longer recordings, and sleep-staged data were not tested.

SEEG · epilepsy · seizure onset zone · aperiodic exponent · spectral power · interictal · AUC · OpenNeuro