Removing contact name groups from the common average reference increased 80 to <150 Hz strong-edge turnover more than matched random removal in 19 SEEG records
We tested whether reference contact coverage changes strong SEEG correlation edges. We used two interictal minutes in each of 19 public records from OpenNeuro ds003876, version 1.0.2. The band was 80 to less than 150 Hz, with 118 to 122 Hz excluded. We removed a contact name group or the same number of random contacts from the common average reference. Both tests used the same samples and evaluation edges. Mean turnover of the strongest 10 percent of absolute correlation edges was 9.803 percent for group removal and 5.410 percent for random removal. The mean difference was 4.393 percentage points. The patient bootstrap 95 percent interval was 3.379 to 5.458 percentage points. 19 of 19 patient differences were positive. Patient independence is assumed. Contact name groups are unverified shaft proxies. This is a reference sensitivity result, with no test of clinical value. The package includes covariance data, exact removal sets, code, source hashes, raw rebuild checks, and an independent primary repeat.
stereo-electroencephalography · common average reference · correlation networks · reference sensitivity · edge turnover · contact coverage
References
4- usesReference choice reduced mean 80 to 150 Hz rank agreement more than time change in 19 public SEEG records sci:sha256:16d6a5e5126b007f6f6be4f042d829df32ccd751f41bcfda4c9b9a3f5a80fce2
This work supplied the record selection and EDF range reader used by the covariance package and the new raw rebuild.
- usesRemoving contact name groups from the common average reference changed 80 to 150 Hz power ranks more than matched random removal in two minutes from each of 19 SEEG records sci:sha256:7d7114c8b7779ef3a3204c20d57fcce812aae64cc640cd41debc7191def95bdd
This work supplied the 19 band covariance files, source metadata, and raw retrieval recipe. We rebuilt all files from new raw downloads and calculated new correlation edge results on the same cohort.
- usesdoi:10.18112/openneuro.ds003876.v1.0.2
Source dataset used for the 19 interictal SEEG records; recorded version 1.0.2 and CC0 metadata.
- doi:10.1093/braincomms%2Ffcae165
Background on re-referencing, connectivity choice, and incomplete spatial sampling in intracranial EEG networks.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesEqual name group reference weights reduced mean beta power change after simulated alternate contact loss in two minutes from each of 19 SEEG records
Its whole name group reference removal result helped define the controlled reference pool intervention. The new test removes contacts within one group and compares two reference weights for beta power. Its correlation outcomes are background, not extra patient evidence.
- usesReducing diagonal load increased strong negative partial correlation density in 80 to less than 150 Hz CAR matrices from 19 SEEG records
This study used its 19 covariance inputs, source code, and fixed retrieval records. These files enabled the minute matrices and the diagonal load audit.
- usesSingle contact checks missed top 10 percent high gamma SEEG edge losses after name group removal from the common average reference in 51 patients
This work supplied covariance data for 19 NIH or JHU patients. It also supplied source selection, retrieval recipes, and the group reference method.