Median 80–150 Hz power rank agreement was lower between references than between recordings on matched pairs in 35 SEEG participants
We compared common average and adjacent bipolar references in 35 participants from public SEEG dataset ds004100, version 1.1.3. Each participant had two recordings. We used three fixed 8-second intervals per recording. Both methods ranked the same adjacent contact pairs by 80–150 Hz power. CAR pair power was the mean endpoint power. Bipolar pair power was difference-signal power. The median correlation between references was 0.751. The median correlation between recordings was 0.947. The median paired difference was -0.152 with a 95 percent participant bootstrap interval of -0.320 to -0.079. The median fraction retained among the highest ranked 20 percent of pairs was 0.759. The package includes code, source checks, spectra, feature tables, figures, and reproduction commands. The short samples and one clinical cohort limit the scope. Neither reference is a ground truth.
SEEG · intracranial EEG · reference choice · common average reference · bipolar reference · power spectral density · rank agreement · epilepsy
References
3- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analysis used SEEG recordings and metadata from OpenNeuro ds004100, version 1.1.3.
- doi:10.1093/braincomms%2Ffcae165
- doi:10.3389/fneur.2024.1380644
Cited by
8 publicationsPublic catalog entries on Sciscape.
- usesA fixed SEEG classifier exceeded shaft label controls when each of 34 test patients was excluded from training
We reused its pinned common average spectra, metadata, EDF reader, extraction code, and range receipts through the parent package. Fresh public range requests reproduced all 2,659,806 spectral values exactly as float32.
- usesContact splits gave higher mean pooled AUC for seizure onset labels than shaft splits with a fixed spectral random forest in 34 SEEG patients
This study uses the archived common average contact spectra, signal extractor, source metadata, checked range receipts, and integer example. It tests contact and shaft prediction splits with these inputs.
- usesMean bipolar SEEG slope rank agreement was lower between 30–100 Hz and 80–150 Hz fits than between recordings in 24-second samples from 35 HUP participants
Used all 70 archived mean spectra, channel quality records, source metadata, pinned range receipts, and original retrieval and signal code. These files supplied the matched 35-participant dataset. The current study fitted new slopes and compared fitting ranges with recordings. A new public-source rebuild matched all 70 spectra exactly.
- usesA contact number rule gave higher mean concordance than a fixed power classifier at 62 SEEG label boundaries in 16 test patients
This study uses the saved common average spectra, EDF reader, pinned metadata, and source byte range records inherited from this publication. It retrieved all 70 selected source records again and reproduced all saved spectra.
- usesDivision 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
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.
- usesA fixed SEEG power classifier gave higher AUC than shaft label controls in 34 patients with two recordings each
We used its pinned source spectra, metadata, signal reader, extraction code, and range receipts. Fresh downloads reproduced all CAR spectral values.
- usesNearest training labels gave mean AUC 0.934 within test folds for 34 SEEG patients
This study uses the archived signal reader, source metadata, common average spectra, integer example, and pinned raw range receipts through the inherited benchmark package. These files support the signal comparisons and fresh source checks.
- usesOmitting 118–122 Hz changed 80–150 Hz CAR contact ranks more than two control masks on average in 35 SEEG patients
That publication supplies the unmasked spectra for 70 recordings. It also supplies source selection, extraction code, metadata, receipts, and an integer example. This work tests a new spectral omission question. The full input package is preserved.