Output phase controls gave excess rejections under an independent contact phase model based on two public SEEG recordings

We tested phase controls for bipolar SEEG signals that shared a contact. We used public resting-state data from two patients in OpenNeuro ds005670, version 1.0.0. The null model preserved each recorded contact spectrum and set contact phases independently. At a 5% lower-tail test level, output phase controls rejected 100.00% and 100.00% of null edge tests in the 1 to less than 45 Hz calibration. Contact phase controls rejected 4.60% and 4.45%. The calibration used spectra from the first 30-second block and 20 shared-contact edges per patient. Each calibration used 100 null samples. A separate analysis used 13 blocks per patient and four frequency bands. The package supplies a covariance formula, a reusable overlap check, control code, derived data, and a pinned retrieval recipe. The result applies to independent recorded-contact phases. It does not establish clinical network accuracy.

stereo-electroencephalography · SEEG · bipolar montage · phase randomization · connectivity · covariance · null model · shared contacts · signal processing