A fixed low gamma correction reduced mean high gamma map error between two 60 second interictal records in 35 SEEG patients under common average reference

We tested whether low gamma changes improve transfer of a high gamma contact map between two interictal records from the same patient. We used public SEEG data from 35 patients in OpenNeuro ds004100 version 1.1.3. Each record supplied 60 seconds. High gamma was 80 to 150 Hz. Low gamma was 30 to 79.5 Hz. Selected line frequency bins were removed. The bin sets did not overlap. Maps used common average reference and centered log power. We added the low gamma map change to the first high gamma map with a fixed coefficient of one. Mean patient squared error fell from 6.277 to 2.559 dB². Mean reduction was 3.718 dB², with a 95 percent patient bootstrap interval of 2.024 to 5.755. Error was lower in 29 of 35 patients. All fixed descriptive checks had positive mean reductions. Fresh raw retrieval checked all 70 records. Separate decoding, spectrum, and scalar calculations passed. All nine numerical result files were reproduced byte for byte. The package preserves code, powers, predictions, random draws, pinned retrieval recipes, and checks. The correction requires low gamma from the target record. The result does not establish a neural cause, seizure onset accuracy, or clinical value. Physical excerpt separation is unknown. The cohort is reused.

stereoelectroencephalography · SEEG · gamma power · power maps · signal transfer · prediction error · common average reference · epilepsy