In eight patients, polarity reversal altered responder-channel CCEP maps, but the planned all-channel test failed

We compared early (10–50 ms) cortico-cortical evoked potential (CCEP) maps at the same bipolar SEEG contact pair and current. The analysis used an archived single-trial table from OpenNeuro ds005448 v1.0.0: eight patients, 90 patient–pair–current units, and 4–5 pulses per direction. Amplitude was each response's projection onto the same-direction normalized mean template; no raw signals were read. We split pulses into even and odd indices and compared within-direction with between-direction map correlations. The planned primary all-channel test did not support the prediction: mean D = −0.065 (95% CI −0.129 to −0.002), 2/8 patients with D > 0, exact sign-flip p = 0.11. The planned Spearman sensitivity test gave mean D = −0.162, 0/8, p = 0.0078. On responder channels (ratio ≥ 3 in either direction; 81 units), the planned sensitivity analysis gave mean D = +0.123 (95% CI 0.064 to 0.187), 8/8, p = 0.0078; mean within- and between-direction correlations were 0.74 and 0.60. Responder sets overlapped partly (median Jaccard index 0.40, including noise). Responder-strength asymmetry was 0.64 between directions versus 0.40 within direction (difference +0.24, 7/8 patients, p = 0.039); in 19% of pairs one direction produced at least twice the total responder amplitude. The direction whose first named contact had the lower number produced a smaller total response (mean log2 ratio −0.19, about 0.88×; 7/8, p = 0.039), but the source-label anode/cathode convention is unknown, so this is not interpreted as an anodal or cathodal effect. A post hoc non-responder analysis found mean D = −0.253, 0/8, p = 0.0078. Projection of noise onto a template made from the same pulses can bias D downward on non-responder channels; therefore the all-channel test is not valid with this amplitude measure, and the bias may make the positive responder result smaller. Limitations are eight patients at one center, 4–5 pulses per direction, clinical current and pair selection, same-pulse responder selection, and stimulation in separate blocks. The measure compares projected amplitudes, not waveform sign, shape, or latency. The responder result was a planned sensitivity analysis; the planned primary test failed for a method reason. Polarity may affect responder-channel maps and total response strength; confirmation needs raw waveforms and cross-validated templates, held-out responder selection, and control of order and time effects.

SEEG · cortico-cortical evoked potentials · stimulation polarity · intracranial electrophysiology · brain-computer interfaces · split-half correlation