Baseline division lowered mean SOZ label AUC within contact name groups for common average gamma power in 30 SEEG patients

We tested whether baseline division improved seizure onset zone (SOZ) contact order in public stereoelectroencephalography (SEEG) data. We used OpenNeuro ds004100 version 1.1.3 and fixed gamma scores from the first ten seconds after annotated onset. The primary score used a common average reference and 30 to 80 Hz power, with 55 to 65 Hz omitted. We compared SOZ and other contacts within contact name groups. In 30 patients and 60 seizures, mean patient AUC was 0.836 for seizure power alone and 0.649 after baseline division. The mean change-minus-absolute difference was -0.187. Its 95% patient bootstrap interval was -0.264 to -0.114. The difference was negative in 26 patients. All nine planned descriptive checks gave a negative mean difference. Fresh raw retrieval reproduced all saved arrays. An independent reader checked all 62 source recordings. A fresh package copy reproduced all 17 numerical analysis files byte for byte. The package preserves code, power arrays, metadata, raw retrieval recipes, results, and instructions. The result measures agreement with source clinical labels. It does not establish a neural cause or clinical benefit. The cohort was reused from prior work. Patient inference assumes independent patients and does not include label or onset error.

stereoelectroencephalography · seizure onset zone · gamma power · baseline normalization · area under the curve · contact ranking · common average reference · epilepsy