Contact medians changed 20 to 500 ms bipolar SEEG templates under two voltage references at 83 sites in eight people
We tested the order of contact subtraction and a median across five calibration pulses. We used public SEEG data from OpenNeuro ds004696 version 1.0.1. A common voltage reference changed the contact-median template at 83 of 83 sites in eight people. The mean template RMS change was 42.34% of calibration bipolar RMS. The descriptive 95% subject interval was 33.60 to 50.87%. The recorded bipolar traces stayed fixed. Subtracting contacts within each pulse before the median kept the template fixed. The mean subject change in later-pulse error for the new reference was -9.49%, with interval -16.75 to -4.38%. Reference invariance does not by itself show lower prediction error. The package preserves exact voltage windows, a proof, source hashes, all scores, checks, code, and four figures. This is an offline method test in a reused epilepsy cohort. It does not show better brain-computer interface control.
stereoelectroencephalography · signal processing · electrical stimulation · voltage reference · median template · brain-computer interface
References
6- usesAn offline median reference created 70–110 Hz voltage from 1–30 Hz SEEG inputs in eight people sci:sha256:404621c2f52f9069133facf722ec222cc635acccf0730258cb13e16a7d9e2181
Used the earlier nonlinear spatial-reference test as related method context. The present test uses a median across pulses at each contact and a different invariant input constraint.
- usesGeometric medians of five traces increased 20–500 ms SEEG error compared with means in eight sessions without added faults sci:sha256:1cb656a992dd0eb1b1c33903e515a44b6dc4db088d73ffeb8232f697aed7aa05
Used the prior five-trace robust-template study to define the distinct operation-order question and to state that reference invariance and clean prediction error are separate properties.
- usesA background model lowered mean voltage error with an early reference and increased mean shape error at 20 to 100 ms after five fit pulses at 86 sites in eight HAPwave SEEG sessions sci:sha256:8fb26571079cd151b4ee7e9f54e746a41971223ba27c94bcbf67ee008c4815fb
Used the pinned source manifest, source metadata, all eligible candidate pair tables, and GPL source download recipe. Selected new contacts by labels and extracted A, B, and C from complete verified sources.
- usesRecorded pulse gaps from 2 to 8 seconds had no clear mean SEEG gain relation in eight sessions with site and order controls sci:sha256:77ca2cd94f79de36a61ae64814637757c4181c890e0e63769f25e428fe358ac5
Used the fixed-current source event mapping and checked its extraction code to map good event rows to MATLAB trial rows.
- usesdoi:10.18112/openneuro.ds004696.v1.0.1
The study analyzed source voltage data and metadata from OpenNeuro ds004696 version 1.0.1.
- usesdoi:10.1523/JNEUROSCI.2201-22.2023
This is the source experiment paper for the OpenNeuro recordings analyzed here.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesPeaks from five SEEG pulses did not show a clear mean timing error decrease over a shared patient time at 20 to 100 ms in 83 cases from eight people
Used its exact float64 voltage windows, 83 contact choices, pulse identities, pinned source hashes, and GPL retrieval and selection code. These inputs enabled the new peak prediction comparison without another full download of all eight source files.
- usesSample medians across five pulses created 70 to 110 Hz power after 4 to 30 Hz Fourier projection at 83 SEEG sites in eight people
This study uses its exact source windows, 83-site cohort, pulse rows, contact choices, source records, and retrieval code. It tests a median across pulses after a fixed Fourier projection. The new source checks repeat the extraction for all eight people.
- usesCovariance rank checks fitted to five reference SEEG vectors and one query accepted every query in rank-five groups at 83 sites in eight people
This work supplies exact voltage windows, 83 metadata-selected readouts, pulse roles, source metadata, hashes, and retrieval code. We retain those inputs and test covariance rank scores.