Class difference subtraction kept 99.99% of added train rate signal energy through fit and test in 20 SEEG records with fixed edges
We tested pulse template subtraction on public SEEG data from 20 people. We added known sine and cosine signals before template fit, with edge times fixed. At the nominal train rate, class difference subtraction kept a mean of 99.9906% of the added waveform energy; full subtraction kept 0.3531%. Both methods reduced early recorded voltage energy by about 74%. This voltage reduction does not prove that artifact was removed or that neural activity was kept. The test shows a constraint for brain interfaces that read during stimulation: a template can learn a wanted repeat signal and then remove it. A signal added only after template fit will miss this loss. The package provides a signal retention audit, tested code, public data epochs, and a proof for an ideal clock. It does not provide a validated neural recovery method.
stereo electroencephalography · electrical stimulation · signal subtraction · signal retention · template fitting · brain-computer interfaces
References
4- usesAlternating templates fitted to 16 pulses reduced SEEG variation at −1 to 2 ms but did not establish a mean reduction at 3 to 15 ms in 20 recordings sci:sha256:73d0d13dcaaf20c4458cadb4703df25525ae115101aa7d7ded22ca14823e54d6
This work already proved additive preservation identities for shared and alternating waveforms. We read it after our main scores and used it to identify prior results. The new audit extends those identities to continuous frequencies, actual edge times, timing leakage, and nearby frequency amplification.
- usesRepeatable pulse class differences exceeded baseline at 3 to 15 ms after detected edges in 17 of 20 SEEG recordings sci:sha256:879b5223cbf14d14e8e6ed3642b3968e076bf8d799c75c3c86fea26e328c4ab9
We used its 20 SEEG epoch files, edge indices, channel selection, source manifest, and retrieval and extraction code. We rebuilt every epoch from public source files before the new frequency audit.
- usesdoi:10.18112/openneuro.ds005169.v1.0.0
The analyzed SEEG epochs come from OpenNeuro ds005169 version 1.0.0; its recorded license is CC0.
- doi:10.1002/hbm.25749
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesLonger interpolation gaps raised mean error for known 70 to 150 Hz signals in 20 SEEG records with fixed edges
We used its 20 SEEG epoch files, source manifest, and template, extraction, and retrieval code. We rebuilt all epochs before the gap comparison.
- usesA rolling 12-pair mean gave higher contrast prediction error than a fixed 12-pair mean at 3 to 15 ms in 14 of 20 early SEEG trains
Its prior fixed-template common-signal result informed the preservation test. We extend the check to a mean that updates after each completed pair, with fixed edges and equal additions to both pulses.