A spatial filter with one direction reduced a later pulse class difference measure in 19 of 20 SEEG trains
We tested a small filter for recording during brain stimulation. We used one train from each of 20 people in OpenNeuro ds005169, version 1.0.0, with 1,671 stored bipolar SEEG channels. The first 24 pulses supplied the filter weights. The next 24 pulses supplied two separate score blocks. One spatial direction was learned from the pulse class difference at -1 to 2 ms after detected voltage edges. It reduced the signed, baseline-subtracted class difference measure at 3 to 15 ms in 19 people. The mean removal was 3.56 µV², with a descriptive 95 percent interval of 1.84 to 5.66. Removal exceeded the exact expectation for a random direction by 3.47 µV², with an interval of 1.79 to 5.53. The filter removed 46.7 percent of the group class difference measure and 10.3 percent of the shared response measure. These are ratios of signed repeatable products, not fractions of neural activity or artifact. The source states that pulse polarity alternates, but it has no individual hardware pulse markers. The data cannot establish clean neural voltage or better decoder performance. This is an exploratory reuse of a prior benchmark. The package contains voltage epochs, fitted directions, figures, all person results, source hashes, code, and reproduction instructions. A full source rebuild reproduced all 20 epoch files and all 24 numerical result and model files byte for byte.
intracranial EEG · spatial filtering · brain-computer interface · electrical stimulation · pulse class · exploratory analysis
References
4- 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 unfiltered pulse and baseline epochs, source channel selection, detected edges, extraction code, fixed metadata, and source hashes. We added a filter learned on the first 24 pulses and scored it only in two blocks from the last 24 pulses.
- usesdoi:10.18112/openneuro.ds005169.v1.0.0
Source voltage and metadata were drawn from OpenNeuro ds005169 version 1.0.0; the saved dataset metadata records this version-specific DOI and CC0 license.
- supportsdoi:10.1002/hbm.25749
Describes the source cortical stimulation method that uses alternating polarity to help separate stimulation artifacts and responses.
- doi:10.1007/BF02534144
Background reference for spatial projection methods.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesEarly pulse projection reduced 20–100 ms site accuracy in two of four offline SEEG recordings
Its one-direction spatial filter motivated the early voltage projection question. This study uses a different SEEG dataset and tests withheld site accuracy against matched random dimension removal. It does not reproduce the earlier later pulse class difference endpoint.
- usesA baseline covariance filter did not show lower mean squared signal change in 20 short SEEG tests
We use the fixed 20-person ds005169 epochs, source hashes, and CC0 extraction code. We keep its calibration direction and later pulse score. We add a baseline covariance rule and a held-out baseline signal change test. Independent matrix checks reproduce the original and orthogonal scores. The cohort and pulse blocks are reused.