The polarity-invariant response did not decrease more than the polarity-flipping component in 12 alternating-polarity SEEG trains

We tested whether the polarity-invariant component (S, assumed neural) decreases more than the polarity-flipping component (D, assumed artifact) during alternating-polarity stimulation. We applied a pre-analysis plan to 12 trains at 43.2 Hz from 7 patients in OpenNeuro ds006519 v1.0.0, calculating RMS 2–15 ms after each pulse in the first and last second. The patient-level median-channel contrast log(S_last/S_first) − log(D_last/D_first) was +0.011 (4/7 patients negative; exact two-sided sign-flip p = 0.84). S changed by −0.054 (p = 0.38); D decreased by −0.065 in all 7 patients (p = 0.016), with most of the decrease from 2 patients. A decrease in S after selecting channels with strong first-second S mostly disappeared under a symmetric selection, consistent with regression to the mean. Artifact cancellation in S was partial (train median S/D 0.32–1.12), there was no noise-floor estimate for 3 trains, and the parity-based polarity assignment is an assumption. This analysis includes 7 patients from one centre and protocol; ds005169 was not analyzed.

stereo-EEG · intracranial stimulation · alternating polarity · stimulation artifact · evoked response · adaptation · signal decomposition · OpenNeuro