Eight selected streams reduced mean SEEG power reconstruction error after new session calibration compared with even spacing in eight people
We tested sparse signal transmission with public SEEG data from eight people. Each person had two sessions. We selected eight power streams with first-session calibration data. A new 120-second calibration fit the reconstruction model in session two. Even spacing selected eight comparison streams. Both methods used the same later test samples. Mean full vector normalized error was 0.9112 for selection and 0.9755 for even spacing. This was a 6.6% reduction. Seven people had lower error. The paired mean gain was 0.0643. Its descriptive 95% person bootstrap interval was 0.0179 to 0.1164. The sign-flip probability was 0.0391. Most mean gain came from which streams were transmitted directly. The extra gain from prediction of missing streams was uncertain. The mean fraction of reference error removed in omitted streams was 11.3%. The source has no trial labels and an unresolved voltage scale. The result concerns numerical power reconstruction. It does not show better intention decoding. The package contains source code, data, figures, source hashes, and reproduction instructions.
stereo-electroencephalography · SEEG · power reconstruction · sparse signal transmission · channel selection · brain-computer interfaces · cross-session calibration
References
3- usesA 120-second median calibration left a mean high gamma rate error of 5.1 percentage points for a 10% target in eight SEEG session pairs sci:sha256:878731c694843c59980d6304a735b1a04b5ce68504f08fef057fdf9c73485346
We use unchanged compact power features, raw witnesses, extraction code, and pinned source range records for all eight people with repeated SEEG sessions. The bank selection, reconstruction models, and scores are new.
- usesdoi:10.18112/openneuro.ds005624.v1.0.0
Source data are from OpenNeuro ds005624 version 1.0.0.
- doi:10.1109/TNSRE.2024.3364752
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesOld slopes reduced mean SEEG power error compared with new ridge fits at penalty 10 after 30 calibration seconds in eight people
We used unchanged SEEG feature and witness files, source wrappers and provenance. Its 120-second sparse-reader results informed the new calibration-length question. We added fixed-target scores and an equal-penalty control.
- usesTwo past seconds increased mean unsent SEEG power error by 5.8 percent for linear readers in eight people with eight sent streams and 120 calibration seconds
Its sparse power reader and eight-stream bank comparison on the same recordings informed the new history question. The present result is a new analysis of reused data, not independent confirmation.