In 35 SEEG patients, broadband power and line length separated seizure-onset-zone contacts; the 30–45 Hz exponent did not

We compared simple interictal spectral features of adjacent bipolar stereo-EEG (SEEG) contacts with clinical seizure-onset-zone (SOZ) labels in 35 patients from OpenNeuro ds004100. The analysis used the first 120 seconds of run-01 and included 3,645 bipolar pairs. Per-patient discrimination was assessed with AUCs, sign-flip permutation tests, Holm correction across 10 features, and a normative location correction based on nearby non-SOZ, non-resected contacts from other patients in MNI space. Broadband power (mean AUC 0.686) and line length (0.657) separated SOZ pairs; both remained significant after location correction (0.649 and 0.611). The 30–45 Hz aperiodic exponent did not separate SOZ pairs (raw AUC 0.503); location correction raised its AUC to 0.552, which was still not different from chance. Relative alpha power was lower in SOZ pairs before correction, but the effect diminished and was not significant after correction. Single-feature effects were modest and are not sufficient alone for clinical localization. Limits include one 120-second segment per patient, uncontrolled vigilance and spike content, clinical SOZ labels, coarse MNI-only correction, and simple log-log slope fits.

stereo-EEG · SEEG · epilepsy · seizure-onset zone · interictal · spectral power · aperiodic exponent · line length · AUC · MNI