In 60 s of interictal SEEG from 32 HUP patients, the aperiodic exponent added no SOZ information beyond broadband power; delta power had mean AUC 0.72

Question: does the interictal aperiodic (1/f) exponent separate clinically marked seizure onset zone (SOZ) SEEG contacts from other contacts in the same patient? Data: OpenNeuro ds004100, 32 SEEG patients, the first 60 s of interictal run-01, 3344 bipolar pairs on the same shaft, 357 of them SOZ. Method: Welch PSD, exponent fits in 30–50 Hz and 2–40 Hz (robust, no knee), band powers, and line length. We calculated an AUC for each patient and tested the AUCs across patients with a sign-flip permutation test and Benjamini–Hochberg correction. The 30–50 Hz exponent was at chance (mean AUC 0.47, p=0.48). The 2–40 Hz exponent was weakly above chance (0.59, p=0.02), but the effect stopped after broadband power was removed from each pair in each patient (0.53, p=0.43). Absolute delta power (0.72) and broadband power (0.70) separated SOZ pairs best, and relative high-gamma power was lower in SOZ (0.34). The pattern was the same in Engel 1 and Engel 2–4 patients. Limits: one short clip per patient, no sleep/wake control, no regional normalisation, simplified exponent fits, and six of 38 SEEG subjects not analysed. The package contains numpy-only code, channel features, per-patient AUCs, and SHA-256 hashes of fetched data.

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