Contact group corrections reduced mean theta to high gamma ratio error compared with shaft corrections in paired 60 second SEEG excerpts from 36 patients

We tested a correction for each contact group in public interictal SEEG data. We used paired 60 second excerpts from 36 patients in OpenNeuro ds004100 version 1.1.3. There were 2703 four contact groups on 455 inferred shafts. Corrections used file 01 only. Predictions used narrow ratios in file 02 and were scored against measured wide ratios in file 02. Mean absolute theta to high gamma ratio error was 1.279 dB for group corrections and 2.248 dB for shaft corrections. The mean reduction was 0.969 dB. The 95 percent patient bootstrap interval was 0.793 to 1.145 dB. A population interpretation requires independent and exchangeable patient effects. There were 32 positive effects, one negative effect, and three zero effects. All seven planned checks had positive mean reductions. A secondary spectral decomposition assigned 86.6 percent of the mean span difference to cross terms. Shared contacts and noise can create these terms. They do not identify neural coupling. All 36 compact patient files matched a fresh extraction from 72 pinned public EDF prefixes. A separate analysis reproduced all numerical outputs. This test concerns recorded montage ratios in the same patients. It does not establish transfer to new patients or long time gaps. The package preserves source code, compact spectra, raw retrieval locks, metadata, all scores, bootstrap draws, validation, and reproduction instructions.

stereoelectroencephalography · bipolar recordings · spectral power ratio · theta · high gamma · contact group correction · signal processing