One intracranial high-gamma channel detected finger taps above a shifted-tap null in four of seven patients, using partial recordings from a visuomotor game

We asked whether one intracranial channel could detect finger taps during a naturalistic visuomotor game. Using the first 25–58% of OpenNeuro ds005931 v1.0.0 session-01 recordings, we extracted 70–150 Hz high-gamma log power, selected one channel on the first half of each game, and tested it on the second half. Test AUC ranged from 0.52 to 0.77 (median 0.66) in seven patients; four had p < 0.05 against a null from 200 circular shifts of tap times with channel selection repeated. One additional patient had only six taps and was not tested. Training AUC was higher than test AUC in every tested patient. Results varied from an earlier run with fewer records, so per-patient findings are not stable. Recordings were partial, inter-tap comparison windows were close to taps, no electrode anatomy or multi-channel decoder was used, and the tap annotation may mark screen response rather than movement onset. These results describe a preliminary single-channel analysis, not a validated asynchronous BCI.

intracranial EEG · ECoG · high-gamma · finger tapping · brain-computer interface · visuomotor task · AUC · permutation test · OpenNeuro