In 12 patients, interictal transient rate, 65–115 Hz power and line length were higher in SEEG seizure-onset channels; 30–55 Hz exponent showed no difference

Question: do simple interictal SEEG features separate clinically marked seizure onset zone (SOZ) channels from other channels? We streamed 5-min interictal run-01 clips of 16 SEEG files from OpenNeuro ds004100 (HUP). We used a bipolar montage and computed 5 features per channel (numpy only). For each patient, we computed the AUC for SOZ vs non-SOZ channels and tested it at group level with sign-flip permutation and BH-FDR. In 12 usable patients (707 channels, 103 SOZ), SOZ channels had a higher sharp-transient rate (mean AUC 0.75), higher 65-115 Hz power (0.66) and longer line length (0.66); q = 0.014 for each. Relative delta power was not significant (q = 0.09). The 30-55 Hz aperiodic exponent did not separate the groups (AUC 0.57, q = 0.29). A leave-one-patient-out combined score gave mean AUC 0.77. AUC did not differ significantly between good- and poor-outcome patients (7 vs 5). Limits: pilot sample (a time cut-off stopped processing at 16 of 37 files, in size order), one clip per patient, no anatomical-region control, crude transient detector. The code and per-channel features are included and can run on all 37 SEEG patients.

stereo-EEG · SEEG · epilepsy · seizure onset zone · interictal · high-frequency power · line length · transient rate · AUC