In 44 ds005169 SEEG stimulation trains, a step-sign polarity rule at 4096 Hz found clear pulse-polarity alternation in only 16 trains, so polarity labels must be validated before an S/D artifact test

We checked pulse detection and pulse polarity before a pre-registered test of the alternating-polarity method (S = polarity-invariant part, D = polarity-flipping part). We fetched the train windows of 44 of 112 ds005169 trains (sub-01 to sub-11) and 1 ds006519 43 Hz train. On the channel with the largest jumps, the pre-registered polarity rule gave an alternation fraction of 0.9 or more in 16 trains, 0.1 or less in 1 train, and a mixed value in 27 trains. When we used only pulses with normal intervals, the counts did not change. A principal-component waveform rule gave 17, 1 and 26. Pulses were missed in 15 trains. The ds006519 train gave 0.48. Thus, the polarity labels are not reliable with this rule. We did not run the S/D outcome test, and we did not calculate S or D. The plan, code, provenance and per-train diagnostics are in the package.

intracranial EEG · SEEG · electrical stimulation · pulse polarity · pulse detection · neurophysiology · artifact · reproducibility