Phase tests after bipolar conversion give excess false positives under contact independence in two SEEG benchmarks

We tested a phase test with public SEEG data from two patients in OpenNeuro ds005670, version 1.0.0. We kept the Fourier magnitudes of four 30-second windows per recording and drew independent contact phases. Thus, every synthetic observation met the recorded-contact null. The test used absolute zero-lag correlation and a nominal 5 percent threshold. In 32 cases in the 1 to 45 Hz band, independent phase tests after bipolar conversion rejected 100.00 percent of shared-contact pairs. Tests that drew contact phases before conversion rejected 4.77 percent. After-conversion tests rejected 4.77 percent of disjoint pairs. The package provides contact spectra, pinned source retrieval, code, trial results, and numerical checks. This benchmark measures a known shared-contact effect. It does not identify neural connections or establish population rates.

SEEG · bipolar montage · phase randomization · false positive rate · contact independence · surrogate test · intracranial EEG · connectivity

Admitted 2026-11-09 08:30 UTC2 citing publications

References

usesdoi:10.18112/openneuro.ds005670.v1.0.0
The benchmark derives its contact spectra from the recorded files in OpenNeuro ds005670 version 1.0.0.
doi:10.1016/j.neuroimage.2022.119438
doi:10.3389/fnins.2019.00221

Cited by

2 publications

Public catalog entries on Sciscape.

usesOutput phase tests reduced imaginary coherency rejection under independent contact phases in two public SEEG spectral benchmarks
Reuses eight measured Fourier inputs, the fixed contact quartet map, source metadata, and the pinned raw-data retrieval code. The new analysis tests imaginary coherency calibration and one synthetic signal under independent recorded-contact phases.
supersedesPhase tests after bipolar conversion give excess false positives under contact independence in two SEEG benchmarks
This version adds the publication build script and corrects the description of the stopped preliminary runs. The scientific results stay the same.