Band-power changes after 50 Hz SEEG stimulation varied across patients; 2-46 Hz changes near the site lasted at least 5 s in 4 of 8 patients

Question: how long after a 50 Hz SEEG stimulation train, and how far from the stimulated pair, does band power stay different from baseline? This sets the read-after-write delay for a closed-loop BCI. Data: 17 of 35 recordings (9 patients) from OpenNeuro ds006519. Each recording had one train of up to 5 s at 0.75-3 mA that caused a negative motor response. Method: Welch band power in 1 s bins. We compared each post bin with the mean baseline, against a leave-one-out baseline null (95th percentile). We measured distance from the stimulated pair. Results: pooled over channels, channels less than 20 mm from the stimulated pair exceeded the null threshold in 11-30% of cases in the 2-46 Hz bands for all 5 post bins. The null rate was about 6%, and channels more than 60 mm away gave 6-17%. At subject level, the near effect was not consistent: near was higher than far in 4 of 8 patients (Wilcoxon p = 0.38). The all-channel exceed fraction against the null gave p = 0.074. No band x time cell passed Holm correction (with 9 subjects, this is not possible over 35 tests). Limits: half the recordings, no sham control, and the first post bin starts 0.08-0.84 s after the artifact. All sites were negative-motor sites. Conclusion: a BCI should measure post-train recovery per patient and per site. It should not use a fixed delay. The package contains the code, the features, the figures, and the instructions to reproduce the work.

stereoelectroencephalography · SEEG · brain-computer interface · electrical stimulation · band power · intracranial EEG · motor response · post-stimulation recovery · distance · OpenNeuro