Treated channels were more abnormal, but this did not distinguish seizure outcome in 43 SEEG patients

We tested whether interictal SEEG abnormality was more concentrated in treated tissue among seizure-free patients than among patients who were not seizure-free. Using archived per-channel features from the first 120 seconds of one interictal clip per patient and outcome labels from HUP (OpenNeuro ds004100) and ds003876, we included 43 patients from four centers (24 seizure-free, 19 not seizure-free). The composite D_RS statistic was the Mann–Whitney AUC comparing treated with spared channels. Treated channels were more abnormal in most patients (mean D_RS 0.635; 32/43 above 0.5; sign-flip p = 9×10⁻⁵). The primary outcome comparison did not support the hypothesis: mean D_RS was 0.598 in seizure-free and 0.681 in not-seizure-free patients (difference −0.084, 95% CI −0.194 to 0.023; stratified permutation p = 0.156; outcome AUC 0.39, 95% CI 0.23 to 0.56). Each cohort and each single feature had the same direction, with no significant outcome test. SOZ-AUC also did not differ by outcome. An exploratory control found a larger treated-channel fraction in seizure-free patients, which could lower D_RS. The study had low power, used short clips, pooled ablation and resection labels, and did not correct for treated volume. These results do not exclude small effects and do not test full-length recordings.

SEEG · epilepsy surgery · interictal activity · surgical outcome · resection · ablation · permutation test · OpenNeuro