Removal of five fixed frequency bins changed 70–110 Hz power shifts by a mean 0.337 dB in 21 SEEG patients

We tested a frequency selection rule for brain signal reading after stimulation. We used public OpenNeuro ds006519 version 1.0.0 and saved spectra from a prior sci study. The test had 35 bipolar SEEG records from 21 patients. We compared all 70–110 Hz bins with a rule that removed five bins near 86.4 and 100 Hz. The score was the mean absolute change in the measured power shift across two window pairs. We took the channel median per record and gave each patient equal weight. The mean score was 0.337 dB. Its descriptive 95 percent patient interval was 0.316 to 0.362 dB. This interval was below a chosen 0.5 dB reference. Wider exclusions gave 0.591 dB. The primary rule reversed the sign of a change in at least one window for a mean 13.1 percent of channels. Few reversals had an all-bin change above 1 dB. An added control removed five randomly selected bins. Its mean score was 0.340 dB. The primary rule did not establish a different mean score from this control. The test measures the effect of frequency selection. It does not identify artifacts or measure decoder accuracy. The package preserves spectra, raw audit samples, one complete source record, code, source hashes, three figures, and a full retrieval recipe.

stereoelectroencephalography · intracranial EEG · brain-computer interface · electrical stimulation · high gamma · frequency selection · spectral power · measurement sensitivity