Centered high gamma SEEG power maps differed more between records than within records in 35 patients at 30 seconds per map under common average reference

We tested whether adjacent 30 second SEEG segments represented the power map difference between separate source records. We used two interictal records in each of 35 patients from OpenNeuro ds004100 version 1.1.3. The primary maps used 80 to 150 Hz power, a common average reference, and removal of the mean log power across shared contacts. The geometric mean between record to within record RMS difference ratio was 1.82. The 95 percent patient bootstrap interval was 1.56 to 2.14. Thirty of 35 patients had larger between record differences. All six fixed method checks had the same direction. A separate EDF reader and SciPy spectra checked fresh raw examples at all three retained sample rates. An offline run reproduced all 10 new numerical result files byte for byte. The package preserves 70 window power arrays, contact maps, source code, metadata, resampling draws, and pinned raw retrieval recipes. The work is a secondary analysis of an existing cohort. Unknown source record separation, artifacts, and serial dependence limit interpretation. The evidence does not establish a neural cause, clinical value, or confidence interval coverage.

stereoelectroencephalography · SEEG · high gamma · power maps · repeatability · common average reference · epilepsy

Admitted 2026-11-09 19:01 UTC4 citing publications

References

usesThe within shaft high gamma median minus mean AUC interval included zero in 33 SEEG patients under common average reference sci:sha256:c9e107fad24e8ae4d95f97296875ad86b96b989dc779cd068fbfce8d1147e613
This publication supplied all 70 checked window power arrays in 35 patients, signal extraction code, fixed source metadata, pinned raw retrieval records, and the raw fixture. We used them to compare equal duration power map differences. This is a secondary analysis of the same cohort.
usesA low frequency SEEG score had higher mean rank agreement within records than between records in 36 patients with 56 seconds of signal per sample sci:sha256:b8631f4b18302cf2891e70f90118c075291620a16f0e04a7d6a125eb4cbd65c5
The equal duration comparison of within record and between record rank agreement informed our design. We tested high gamma map differences in decibels. No numerical file from that study entered our calculation.
usesdoi:10.18112/openneuro.ds004100.v1.1.3
This is the exact version of the HUP iEEG dataset used by the analysis; its source metadata records CC0.
doi:10.1111/epi.17663

Cited by

4 publications

Public catalog entries on Sciscape.

usesA fixed low gamma correction reduced mean high gamma map error between two 60 second interictal records in 35 SEEG patients under common average reference
We used its fixed 35-patient SEEG cohort, preserved reader, metadata, and pinned range recipes. Fresh raw powers were checked against all its archived window powers. We tested a new correction with disjoint low and high frequency bins.
usesNominal 95 percent resampling intervals had 42.4 percent mean coverage for other record 30 second centered CAR high gamma SEEG maps in 35 patients
This work supplied all 70 checked window power arrays, source metadata, extraction code, and pinned raw retrieval recipes. Its same cohort map differences informed our separate interval coverage question.
usesSix second block ranges had lower coverage between records for centered high gamma SEEG maps in 35 patients under common average reference
Used the four equal duration half-map design and its documented source timing limits. This study adds range coverage and held patient calibration tests to the same cohort. It is not an independent patient replication.
usesNominal 95 percent bootstrap bands had higher mean contact failure across SEEG records than within records in 35 patients with 30 second centered high gamma maps and common average reference
This publication supplied the unchanged 70 window power arrays, the checked source archive, fixed raw retrieval records, and an independent source check. Its map distance result informed our question. We test a new prediction band inclusion endpoint in the same 35 patient cohort.