In 21 SEEG patients, resected non-SOZ channels were not more abnormal than non-resected channels

We asked whether interictal abnormalities extend from the clinical seizure-onset zone (SOZ) into resected channels not labelled SOZ (RNS), compared with non-resected non-SOZ channels (NRN). We used archived per-channel feature tables from two AutoSci packages, based on the first 120 s of one interictal clip and within-shaft bipolar pairs: 14 patients from HUP (OpenNeuro ds004100) and 7 from the NIH site in ds003876. Twenty-one patients had at least 3 RNS and 3 NRN channels. Per-patient RNS-versus-NRN AUCs were tested with two-sided sign-flip tests and Holm correction across four features. No primary test passed correction: spike rate mean AUC 0.585 (15/21 higher; Holm p=0.19), 1–40 Hz log power 0.483 (p=0.68), 80–150 Hz log power 0.413 (p=0.19), and line length 0.416 (p=0.15). Power and line-length estimates were lower in RNS; spike rate showed a small, non-significant increase. In secondary comparisons, SOZ versus RNS AUCs were 0.707 for 1–40 Hz power, 0.692 for 80–150 Hz power, and 0.725 for line length (Holm p≤0.010). A same-shaft control in 16 patients gave AUCs of 0.45–0.51, with all Holm p=1. The results indicate that, in these data, interictal power and line-length abnormalities were specific to the clinical SOZ and did not extend to the resection margin outside it. Spike-rate interpretation may be partly circular because clinicians can use spikes to plan resection. Limits include the small sample, one clip per patient, no sleep staging, centre-specific clinical labels, resection labels that may include ablation or be incomplete, and no surgical-outcome split. The archived source feature-table hashes are preserved in the report and results.

SEEG · epilepsy · seizure-onset zone · interictal · resection · spike rate · line length · intracranial EEG · AUC