Heartbeat-locked SEEG signals were at chance level in one patient with a clear ECG, but the test had low power

We audited electrocardiogram (ECG) quality and tested for heartbeat-locked stereo-electroencephalography (SEEG) signals in the first session of 24 subjects in OpenNeuro ds005624. We read 150 s from the middle of each BrainVision float32 recording using HTTP byte ranges. Three subjects had EKG channels; only HUP139 had clear R peaks (199 beats), while HUP136 failed the RR-interval variability criterion and HUP133 had no clear QRS despite passing numeric checks. In HUP139, the fraction of channels with significant heartbeat-locked 1–40 Hz voltage or 70–150 Hz log power (200 random-time surrogate sets, p ≤ 0.01 per channel) ranged from 0 to 0.028 across monopolar, common-average, and bipolar references, near the 0.01 chance level. Median debiased cardiac-locked variance was between −0.002 and 0.000 in all reported cells. Injection tests showed that the procedure detected a 0.2 z heartbeat-locked bump in fewer than 10% of channels in each measure and subject. The null result therefore does not show that cardiac contamination is absent; it only excludes large effects under this limited test. Only 1 of 24 subjects had a clear ECG in the analyzed segment. Larger samples and recordings with clean ECG channels are needed. The source reports unresolved differences in channel counts between analysis scripts and assumes 60 Hz line frequency.

stereo-electroencephalography · heartbeat-locked signals · electrocardiogram · cardiac contamination · signal quality · surrogate test · OpenNeuro