Early ictal high-frequency power marks clinical seizure-onset contacts but treatment coverage does not predict outcome in 35 SEEG patients

In OpenNeuro ds004100, early ictal 30–100 Hz power separated clinically marked seizure-onset-zone (SOZ) bipolar pairs from other pairs in 35 SEEG patients with labels (mean per-patient AUC 0.78; 86% of patients above 0.5; Holm-adjusted p = 0.0002). Pre-ictal delta power had mean AUC 0.71, an Epileptogenicity-Index-like score 0.66, and the 1–12 Hz ratio 0.53 (p = 0.44). In the primary outcome analysis (20 seizure-free and 15 not seizure-free), treatment covered a mean 0.50 versus 0.44 of each patient's top three data-driven onset pairs; the S-versus-F AUC was 0.53 (permutation p = 0.73). Clinical SOZ coverage also did not separate outcome (AUC 0.56, p = 0.55). SOZ agreement is partly circular because clinicians mark SOZ from ictal recordings. The analysis used one seizure per patient, a 70-s window, and contact-level treatment labels; it had low power for small outcome effects and no artifact rejection or anatomical control. Files were retrieved from live OpenNeuro URLs accessed 2026-10-10; the exact dataset release is not established in the archived fetch records.

SEEG · epilepsy · seizure onset zone · ictal power · high-frequency activity · surgical outcome · OpenNeuro ds004100