Responsive channels in ds005169 showed a 28% greater decrease in polarity-invariant response than artifact (p=0.006); all-channel tests were not significant

We tested whether the polarity-invariant component (S) of pulse-locked responses during alternating-polarity bipolar SEEG stimulation adapts over a train, while the polarity-flipping component (D) remains stable. We analyzed 8 trains in 5 patients from OpenNeuro ds006519 (primary) and 33 trains in 13 patients from ds005169 (replication), without pooling datasets because patients may overlap. The primary all-channel contrast, median log(S_last/S_first) minus median log(D_last/D_first), did not reach p<0.05 in either dataset (ds006519 ratio 0.95, p=0.56; ds005169 ratio 0.96, p=0.063). In the secondary analysis of channels with first-second S greater than three times the noise floor, the contrast ratio was 0.77 in ds006519 (4/5 patients negative, p=0.125) and 0.72 in ds005169 (12/13 negative, p=0.0063); S alone in ds005169 had ratio 0.68 (p=0.00049). D remained near constant, and pre-stimulation null contrasts were not significant. This supports an adapting neural component in S but does not show that S is free of artifact. Samples were small, many trains were excluded for failing alternation or duration criteria, and selection of responsive channels by S_first may cause regression to the mean; the D contrast partly, but not fully, controls this. The 2–15 ms window may contain residual artifact.

intracranial EEG · SEEG · alternating-polarity stimulation · stimulation artifact · neural adaptation · pulse response · sign-flip test