Removing the largest contact group from the average reference replaced 14 percent of strong beta imaginary coherency links in 60-second samples from 35 SEEG patients

We tested reference sensitivity in two 60-second public SEEG samples from each of 35 patients with epilepsy. Within each file, source voltages and candidate links stayed fixed. We removed the largest named contact group from the uniform average reference. The strongest 10 percent of beta imaginary coherency links had mean replacement of 14.00 percent. The 95 percent patient bootstrap interval was 11.35 to 17.11 percent. Mean replacement between files was 80.20 percent under the full reference. The paired mean contrast was -66.21 percentage points (95 percent interval: -69.74 to -62.36). Reference replacement was smaller in all 35 patients. Population use of the intervals requires independent and exchangeable patients. File replacement includes estimator noise and unknown recording differences. The results do not identify true neural links or verify physical shaft identity. All independent numerical checks passed. A second extraction and analysis gave identical files. The package preserves code, scores, selections, results, source hashes, a raw example, and a public retrieval recipe.

stereoelectroencephalography · SEEG · imaginary coherency · average reference · contact group · functional connectivity · epilepsy · reference sensitivity