CAR and Laplacian references changed relative gamma ranks more than short time shifts in 36 HUP SEEG patients

We tested reference choice against short time changes in public SEEG data from 36 HUP epilepsy patients. Each patient had two interictal runs and three 30-second windows per run. For relative gamma power, median rank agreement was 0.511 across common average and local Laplacian references. It was 0.916 across windows within a reference. The median patient difference was 0.407 (95% patient bootstrap interval 0.313 to 0.468). The difference was positive in 36 of 36 patients. The package provides contact scores, exact neighbor maps, source metadata, pinned retrieval instructions, source code, and patient results. The result applies to the tested short windows and reference pair. It does not show which reference is better or establish clinical accuracy.

stereo-EEG · SEEG · common average reference · local Laplacian reference · relative gamma power · rank stability · epilepsy