Ictal Epileptogenicity Index identified SOZ channels but did not distinguish surgical outcome in 35 HUP SEEG patients

In 35 patients from the HUP SEEG dataset (OpenNeuro ds004100 v1.1.3), the ictal Epileptogenicity Index (EI) identified clinician-labelled seizure-onset-zone (SOZ) channels in two seizures (mean within-patient AUC 0.720 in run-01 and 0.719 in run-02). EI was not more discriminative than pre-ictal 1–97 Hz power, while its run-01 advantage over ictal gamma gain did not replicate. EI improved on onset latency alone in both seizures. It did not distinguish Engel I from Engel II–IV surgical outcomes: the run-01 resection AUCs were 0.702 and 0.703 (p = 0.99), with a similar null result in run-02 (p = 0.84). Between-seizure EI channel ranks were moderately stable (median Spearman rho 0.41). The analysis used 60 s around annotated onset, one center, small outcome groups, and clinical SOZ labels partly derived from ictal EEG; no artifact rejection or anatomical control was used. Thus, the observed SOZ agreement may be favoured by label circularity, and the outcome analysis has limited power.

epilepsy · SEEG · ictal EEG · epileptogenicity index · seizure-onset zone · surgical outcome · AUC · intracranial EEG