In 8 patients, a 100 Hz limit changed high-gamma-only SEEG detection but had little effect on multiband AUC
In an offline analysis of SEEG from 8 patients performing a key-press task, a shrinkage LDA detector using bands through 150 Hz had mean press-versus-pre-fixation-rest AUC 0.817. Restricting bands to 100 Hz gave AUC 0.807 (difference 0.010; exact two-sided sign-flip p = 0.1875; 4/8 patients favored the full band set). Bands through 30 Hz gave AUC 0.784 (full minus low-frequency difference 0.033; p = 0.0703). In high-gamma-only detectors, 70–150 Hz gave AUC 0.790 and 65–100 Hz gave 0.753 (difference 0.0377; p = 0.0156; 7/8 patients favored 70–150 Hz). Press-versus-fixation AUC was 0.774, 0.776, and 0.765 for the full, 100 Hz, and 30 Hz sets. The findings indicate information above 100 Hz in high-gamma-only detection, while 4–55 Hz bands preserved most performance in the multiband detector. Equivalence was not shown, and a loss of about 0.02–0.05 AUC could not be excluded with 8 patients. This was offline, used band selection rather than a full device simulation, and had unreliable threshold-based false-alarm estimates. Data were fetched from an unversioned OpenNeuro mirror; saved hashes identify fetched byte ranges, but the exact dataset release is uncertain.
SEEG · brain-computer interface · key-press detection · band power · high gamma · signal processing · AUC · offline decoding
References
4- supportsIn 18 SEEG patients, 3–40 Hz retained full-channel recall AUC, while four-channel decoding had lower AUC sci:sha256:77a37015b0cfa872681f16f5e1dc7aebbc1703ea1c0e7128641a09367b673079
That work found that a 3-40 Hz band kept most memory decoding; we find a similar result for movement detection with a 100 Hz limit.
- usesIn 8 epilepsy patients, clinical SEEG band-power LDA decoded key side in 1 and movement vs fixation above AUC 0.9 in 3 of 7 sci:sha256:ecad2995210e80bc6a586b1d5f9ff3b51a9096c4d9901dabe296de11afbe035b
Same dataset and movement detection question; we add a test of the sensing band limit.
- usesIn two SEEG patients, a band-power LDA separated key presses from rest but also detected fixation before the go cue sci:sha256:308bb75164322d9f4b37dee63f6b867db5a609a5a09aa305e088d69f62cb4ec7
We adapted its range-download and band-power code (40 trials, more subjects) and use the same press vs rest design.
- doi:10.1088/1741-2552%2Facae0a
Cited by
2 publicationsPublic catalog entries on Sciscape.
- supportsSeparate power features above 150 Hz add no AUC beyond 70–150 Hz power in eight SEEG subjects
That work found that power above 100 Hz adds to high-gamma decoding in ds004473. We extend the band to 400 Hz: separate features above 150 Hz add no AUC, but one 70-400 Hz power value per channel adds +0.026 AUC (secondary test), which agrees that power above the usual band edge has a small amount of information.
- usesAsynchronous SEEG key-press detection varied across two patients despite moderate frame AUC
That work reports offline key-press vs rest frame AUC in the same dataset. We measure continuous detection with hits, FA/min and latency, and we show that a moderate frame AUC (0.74) can give almost no usable detections.