Mean 70 to 100 Hz power rank correlation between SEEG shafts exceeded 60 s common phase controls in 36 patients

We tested whether fixed band voltage cross spectra could reproduce power rank correlation between separate SEEG shafts. We used two 60-second public interictal excerpts from each of 36 patients with epilepsy in OpenNeuro ds004100, version 1.1.3. We selected one adjacent bipolar pair per shaft. The mean 70 to 100 Hz power rank correlation in 0.5-second bins was 0.04448. In 199 common phase controls per file, the mean was 0.00209. These controls kept all selected band cross spectra. The mean excess was 0.04240, with a descriptive 95 percent patient bootstrap interval of 0.03592 to 0.04922. All 36 patient means had positive excess. A second control kept mean power in five-second blocks. Its excess was 0.00800, with an interval of -0.00073 to 0.01612. This interval included zero. The results depend on the control time scale and score. They do not establish a direct neural connection. The package keeps code, all band inputs, all control scores, source versions and hashes, metadata, and repeat instructions. All 584 numerical arrays matched a fresh offline run exactly.

SEEG · intracranial EEG · power correlation · surrogate data · phase randomization · high gamma · epilepsy · reproducibility