In 43 SEEG patients, tonic 250–450 Hz power separated clinical SOZ pairs but did not outperform 80–150 Hz power

Question: in interictal NREM SEEG, does tonic power in the fast-ripple band (250-450 Hz) separate clinical seizure-onset-zone (SOZ) bipolar pairs better than power in the ripple band (80-150 Hz)? Data: OpenNeuro ds007118, ds007119 and ds007120 (snapshot 1.0.0), 43 patients, 728 same-shaft bipolar pairs, 160 SOZ pairs, 30-120 s of each first run, 1000 Hz sampling. The selection and the EDF bytes are identical to AutoSci a3d06b3c (all 43 SHA-256 values agree). Method: we wrote the plan before we read signals. We calculated Welch log power in 80-150, 150-250 and 250-450 Hz with line-noise bins removed, and a per-patient AUC of SOZ pairs against other pairs. We used sign-flip tests over patients, bootstrap 95% CIs and Holm correction. Results: 250-450 Hz power separated SOZ pairs (mean AUC 0.670, CI 0.581-0.753, Holm p = 0.0017; 33 of 43 patients above 0.5). 80-150 Hz power gave mean AUC 0.683. The paired difference was -0.013 (CI -0.069 to 0.038, Holm p = 0.71). Thus the data do not show a fast-ripple advantage, and a mean gain above about 0.04 AUC is unlikely. 250-450 Hz power did not add information after 80-150 Hz power (residual AUC 0.533, Holm p = 0.71). Within patients, the two bands had a median rank correlation of 0.80. In the 15 Detroit patients, no band separated SOZ pairs; in the 28 external patients, all bands did. No patient had a steep 250-450 Hz filter roll-off. Limits: this is tonic power, not detected fast-ripple events. At 1000 Hz sampling, the 250-450 Hz band can contain much noise or broadband activity. SOZ labels are clinical. 90 s per patient.

intracranial EEG · seizure-onset zone · fast ripples · ripple-band power · high-frequency oscillations · epilepsy · AUC