Fast ictal power increase marks SOZ pairs, but low-frequency suppression and automatic chirp slope do not

We tested three parts of the epileptogenic-zone fingerprint at clinical seizure onset in HUP stereoelectroencephalography (SEEG). Across 107 seizures from 30 patients, within-shaft bipolar pairs at clinical seizure-onset-zone (SOZ) contacts were compared with other pairs. In the preregistered 0–10 s epoch, the 20–120 Hz power change separated SOZ pairs (patient mean AUC 0.758, 95% CI 0.680–0.832; Holm p = 4.5 × 10⁻⁵). The 3–12 Hz power change did not support the predicted suppression (AUC 0.542, 95% CI 0.458–0.626; Holm p = 0.43); group SOZ power increased after onset. An automatic peak-frequency slope also did not separate SOZ pairs (AUC 0.473, 95% CI 0.432–0.515; Holm p = 0.43). Exploratory 0–20 s results were AUC 0.606 (p = 0.021, uncorrected) for low-frequency power and AUC 0.449 (95% CI 0.403–0.489; p = 0.025, uncorrected) for chirp slope. A chirp-positive flag was present in 39% of non-SOZ pairs. Resected-contact labels showed the same pattern. Clinical onset marks may occur after the first electrographic change, so the 0–10 s epoch may miss an early chirp. The automatic slope is not a visual fingerprint rating, and the preictal-spike component was not tested. Results and conclusions apply to these automatic measures, onset alignment, and this dataset.

epilepsy · stereoelectroencephalography · seizure onset · seizure-onset zone · intracranial EEG · power spectrum · frequency slope · OpenNeuro · ds004100