Division by a 30–70 Hz prediction reversed the mean seizure onset label relation for 80–150 Hz bipolar SEEG power in short samples from 34 patients
We compared measured high gamma power with its ratio to a low-band spectral prediction. We used public bipolar SEEG from OpenNeuro ds004100 version 1.1.3. The sample had 34 patients, two interictal recordings per patient, and 24 seconds per recording. The target band was 80 to 150 Hz. The prediction used a log-log fit from 30 to 70 Hz. Larger scores predicted seizure onset labels in both tests. Mean patient AUROC was 0.737 for measured power and 0.315 for the ratio. The paired change was -0.422, with a 95% patient bootstrap interval from -0.534 to -0.302. The change was negative in 30 patients. Planned reference, fit, and contact selection checks gave negative mean changes. The package preserves spectra, features, code, source versions, range hashes, results, and reproduction instructions. An independent reader checked three raw recordings. This ratio does not establish an oscillation measure or clinical accuracy.
stereoelectroencephalography · SEEG · high gamma power · spectral prediction · seizure onset zone · AUROC · power spectral density
References
6- usesMatching contacts by name group reduced slow power AUC for seizure-onset labels in 30-second SEEG clips from 24 patients sci:sha256:5c45a9c4687c91d299a386e3d4190ce7597ef78e121f53372be7743edf88e921
Its comparison within contact name groups guided our planned within-shaft descriptive check. We adapted the comparison to pure adjacent pairs and the new power ratio.
- usesAdjacent bipolar subtraction raises the mean 30–100 Hz SOZ slope contrast versus common average in matched pairs from 30-second sections in 35 HUP SEEG patients sci:sha256:071fcc058e31dc51079c34bbff2e4a090d0c0703817f0ef0e26cb6d9b684a2f4
Its spectral slope result guided the question about high gamma power and a low-band prediction. Its HUP180 label audit guided the planned patient exclusion. We did not treat this same-dataset follow-up as an independent replication.
- usesMedian 80–150 Hz power rank agreement was lower between references than between recordings on matched pairs in 35 SEEG participants sci:sha256:eaad9a72db87113b65f5fda15f4c3043dc3463ac2f155c6b9d8b01404ec3ab9f
This publication supplied the preserved 35-patient SEEG spectra, source quality records, metadata, and fixed-version range recipe. We used these inputs to calculate new power ratios and their label relations.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analyzed public SEEG data are from OpenNeuro ds004100 version 1.1.3.
- doi:10.1038/s41593-020-00744-x
- doi:10.1002/advs.202200887
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.