The after-to-before ratio of median squared SEEG pulse distances was 1.54 at 4–30 Hz and 100–500 ms after stimulation in eight epilepsy patients

We tested whether a low-band increase in SEEG pulse variation remains when the measure resists one extreme pulse. We used public OpenNeuro ds004696 version 1.0.1 and fixed remote bipolar readouts. The final five good pulses gave 886 pair instances at 86 sites in eight epilepsy patients. The measure was half the median of ten squared Fourier-vector distances. A common response cancels from each distance. At 4–30 Hz and 100–500 ms, the group after-to-before ratio was 1.537. Its patient bootstrap 95% interval was 1.332–1.779. All eight patient ratios exceeded one. At 70–110 Hz, the ratio was 0.988 [0.950, 1.028]. The low-band ratio stayed at 1.169 [1.024, 1.355] after the worst observed pulse omission at each site. A conservative bound for arbitrary pulse replacement was 0.339 and did not certify an increase. The report gives a candidate measure for interface calibration. It does not prove neural origin or live control benefit. The contribution preserves compact inputs, source hashes, full distances, figures, a proof, and verification records. GPL scripts are omitted, which limits a fresh numerical replay from the SCI package. This is an exploratory extension of earlier work on the same cohort.

SEEG · electrical stimulation · pulse variation · robust statistics · Fourier analysis · epilepsy · brain interfaces