Blanking enables 70–150 Hz SEEG readout to recover in about 0.1 s at least 15 mm from stimulation

We measured how long stereo-electroencephalography (SEEG) readout features remain changed after a clinical single pulse, to estimate a pause for a bidirectional brain–computer interface. We analyzed the first SPESclin run for patients STREEF06–13 in OpenNeuro ds005448 v1.0.0: six stimulation pairs per patient, 529 pulses selected, and 457 analyzed after removing 72 pulses within 1 s of an annotated artefact. For each non-stimulated bipolar channel-site, we compared post-pulse 70–150 Hz (high-gamma, HG) and 4–30 Hz (low-frequency, LF) log power with the −1.0 to −0.1 s baseline in 50 ms bins. With linear interpolation over −2 to +10 ms, pooled HG recovery was 0.10 s at 15–30 mm and 30–50 mm, and 0.05 s beyond 50 mm; within 15 mm it was 0.60 s. LF recovery was 0.45, 0.40, 0.25, and 0.00 s across those distance classes. Without blanking, the first 0–50 ms HG bin was significant at 99%, 95%, 80%, and 46% of channel-sites from nearest to farthest; with blanking these fractions were 57%, 15%, 9%, and 4%. At distances below 15 mm, HG power remained about 0.03–0.04 log10 units below baseline after 100 ms until about 0.6 s. In this dataset, HG readout at least 15 mm from the stimulated pair could resume about 100 ms after a pulse if blanked; closer channels and LF features required about 0.25–0.6 s. Findings apply to clinical SPES at about 0.2 Hz, the tested pulse conditions, 50 ms bins, and the chosen chance rule. There was one run and six pairs per patient, with about 10 pulses per pair; small channel-site effects may be missed. Distances used electrode coordinates without anatomy. The full method, per-patient ranges, results, code, source provenance, and other limitations are given in the main text and supporting files.

stereo-electroencephalography · SEEG · single-pulse electrical stimulation · SPES · brain–computer interface · high-gamma power · low-frequency power · blanking · recovery time · epilepsy