Patient median beta score changes were below 0.006 after removal of one shaft from the average reference in four fixed SEEG records

We tested removal of one shaft from an average SEEG reference. The test used 48 pre-pulse epochs from four patients in OpenNeuro ds004977 version 1.2.0. Surviving raw voltages stayed fixed. The beta score was mean absolute imaginary coherency at 16 to 28 Hz. Each shaft case used its surviving cross-shaft links. Patient medians of mean absolute score change were 0.00276 to 0.00582. Their equal-patient mean was 0.00371. The largest shaft mean change was 0.01517. Three of 54 shaft cases had a link change above 0.05. The largest single link change was 0.1313. The equal-patient mean median absolute 70 to 110 Hz power change was 0.02675 dB. No primary contact comparison exceeded 1 dB. Fixed adjacent bipolar pairs preserved both measures within numerical precision, under the same shared reference and linear processing. Planned checks excluded the pulse shaft and split the epochs. The package supplies an exact spectral update tool, all analysis inputs, raw check samples, trace hashes, source pins, full results, and three figures. All 2,157 raw files and 19 metadata files passed source checks. A separate scalar solver and complete extraction and analysis replays passed. This small audit reuses selected clinical data. It measures a software effect. It does not establish neural coupling truth, task accuracy, a live switch transient, or stimulation benefit.

SEEG · common average reference · imaginary coherency · spectral power · reference change · signal processing · brain-computer interfaces