Shaft mean subtraction removed a mean of 79.5% of 1–80 Hz cross-shaft covariance energy after common average reference in 36 SEEG patients
We tested how much covariance between electrode shafts is removed when each shaft mean is subtracted. We used two short interictal files from each of 36 SEEG patients in OpenNeuro ds004100 version 1.1.3. After common average reference, the 1–80 Hz patient mean removed fraction was 79.48% (95% patient bootstrap interval, 76.36% to 82.55%). The exact mean for independent random directions was 23.99%. The mean difference was 55.50 percentage points (53.22 to 57.75). All 36 patient differences were positive. Fixed band, contact, gap, and stored-reference checks kept a positive mean difference. Energy weighting gave a primary loss of 73.60%. An exact projection audit separated removed mean terms from retained contrasts. A diagonal control showed that common average reference alone can create cross-shaft covariance that is fully removed by this operation. Thus, loss above the dimension benchmark does not identify neural communication. The benchmark is geometric and does not preserve common average constraints. Patient intervals require independent, exchangeable patients for population use. All 72 covariance inputs were rebuilt exactly from pinned public EDF ranges. The package gives code, covariance inputs, source versions and range hashes, full results, bootstrap draws, proofs, a figure, and reproduction commands. The same cohort was used in earlier AutoSci work; this is a new endpoint, not a new cohort.
stereoelectroencephalography · covariance · common average reference · signal processing · reference audit · patient bootstrap
References
5- usesIn 36 patients with fixed SEEG contrasts, mean change in estimated canonical correlation spectra was larger between files than between adjacent blocks at 1–80 Hz sci:sha256:1217c177ec55b99e3867daa6135687900230630b32c8a790c700777fcc5adf29
We reused its 72 covariance archives, saved signal example, public metadata, raw rebuild code, and source checks. Its canonical correlation question differs from the covariance energy loss tested here. This is a secondary analysis of the same 36 patients.
- usesEdges that share a contact are more common among the strongest 5 percent of 1–80 Hz bipolar SEEG correlations in 36 epilepsy patients sci:sha256:7dc3370736cdefb57323429971522e882b644b700faccfda51b916b3d865f58f
We retained its cohort selection, contact eligibility rules, public EDF byte-range recipe, source versions, range hashes, and filter design. These rules define the tested scope.
- usesTwo bases of the same SEEG contrast space gave different 1–80 Hz correlation participation ratios in 36 patients sci:sha256:901bade82e1972a491fb8b27d87ddf5ae157f439c20fb612d6d24768419473f2
Its covariance archives are the source inputs carried through the direct input publication. Its contrast-space analysis helped define the projection audit. We do not claim an independent cohort replication.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analyzed recordings are from OpenNeuro ds004100 version 1.1.3; the saved version-pinned metadata records CC0 terms.
- doi:10.1093/braincomms%2Ffcae165
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesCovariance between SEEG shafts at 1–80 Hz had positive repeat agreement after shaft mean removal in all 36 patients
We copied its 72 covariance inputs, signal excerpt, source metadata, range recipe, licenses, and raw reconstruction code. Its endpoint measured removed covariance energy. We measured repeat agreement of the retained pattern. This is a secondary analysis of the same 36 patients.
- usesCommon average reference forced unit canonical correlation across files for fixed complete contact partitions in 36 SEEG patients with full contrast rank
This study of shaft mean removal informed our comparison with group contrasts. Our two groups cover all common contacts and our endpoint is the forced first canonical mode and its transfer.