In 25 HUP SEEG patients, 150 s of interictal amplitude features identified clinical SOZ channels and remained significant within shafts, but did not differ by surgical outcome

We used OpenNeuro ds004100 v1.1.3 (HUP iEEG Epilepsy Dataset) to test whether simple features from one 150 s interictal stereoelectroencephalography (SEEG) clip separate clinically marked seizure-onset-zone (SOZ) channels from other channels. We analyzed 25 of 38 SEEG patients with SOZ labels: 17 with Engel I outcomes and 8 with Engel II-IV outcomes. Band power, total power, line length, and spike rate were higher in SOZ channels (median per-patient AUC 0.69–0.76; Holm-adjusted p<0.05), while relative high-gamma power was lower (AUC 0.33). In within-shaft comparisons of 21 patients, line length and log total power remained significant (AUC 0.79 for each), but spike rate did not (AUC 0.61). A leave-one-patient-out multivariate model had median AUC 0.81. No feature or multivariate performance differed significantly by surgical outcome. Resection-zone results were weaker; spike rate had AUC 0.71 and within-shaft relative low-gamma power had AUC 0.74. The sample came from one center, included only 25 patients and 8 poor-outcome patients, and used one short clip per patient without sleep/wake or gray/white-matter control. Clinical SOZ labels are not ground truth, and the simple spike detector was not validated against expert markings.

stereoelectroencephalography · SEEG · epilepsy · seizure-onset zone · interictal · signal features · within-shaft analysis · AUC · OpenNeuro