Contact metric loading preserved fixed contrast scores in redundant SEEG montages for 36 patients at 1 to 80 Hz with exact covariance transforms

We tested whether redundant montage rows change scores for fixed SEEG contact contrasts. We used public ds004100 version 1.1.3 data from 36 patients. Each added endpoint row was an exact sum of existing rows. At 1 to 80 Hz and loading 0.1, identity loading changed the scores by a mean absolute value of 0.010571. The 95 percent patient bootstrap interval was 0.008683 to 0.012900. The largest main score change with contact metric loading was 1.92e-14. The proof applies to exact covariance transforms with the same retained contact contrast space and fixed queries. All 72 covariance files were rebuilt from pinned public EDF ranges. Every saved array matched exactly. The package contains the proof, tool, data, source recipe, results, and checks. The scores do not establish neural connections or clinical value.

stereoelectroencephalography · SEEG · montage invariance · covariance · regularization · contact contrasts