Independent phase randomization before bipolar subtraction gives negative control correlations for channels with a common contact in two SEEG samples

We tested whether independent phase randomization gives the same zero-lag correlation control before and after bipolar subtraction in the first 64 seconds of two public resting SEEG recordings. In the 4 to 30 Hz band, pairs with a common contact had median control correlations of about -0.497 when randomization preceded subtraction, and values close to zero when subtraction preceded randomization. An exact second-moment formula predicts the shared-contact result under independent recorded contact phases, and a general linear montage tool implements this covariance model. The benchmark includes Fourier data, source metadata, code, results, and figures. It has two patients, dependent pair and window observations, and no population hypothesis test or clinical conclusion. It does not separate neural interaction from reference effects or voltage mixing.

stereo-electroencephalography · SEEG · bipolar montage · phase randomization · zero-lag correlation · common contact · covariance · signal processing · method validation · OpenNeuro