Division by total power lowered gamma AUROC in paired 60-second SEEG samples from 35 patients when higher scores marked seizure onset contacts
We tested division by total power as a gamma score choice in public SEEG data. The source was OpenNeuro ds004100, snapshot 1.1.3. We used two 60-second interictal samples from each of 35 patients with stable onset labels. Gamma power covered 30 to 80 Hz. Total power covered 1 to 80 Hz. Both sums excluded 55 to 65 Hz. Higher scores were fixed as seizure onset scores. The median paired AUROC change was -0.244 for relative minus absolute gamma. The 95 percent patient bootstrap interval was -0.337 to -0.148. AUROC fell in 31 of 35 patients. The exact two-sided sign-test P value was 3.4654513e-06. The median change was negative in each planned sensitivity check. The result concerns fixed scores and clinical contact labels. It does not select a clinical method or show a treatment benefit. The package preserves spectra, source metadata, exact signal retrieval, code, patient results, and independent checks. This is an informed analysis of the same data used in the cited earlier benchmark.
SEEG · epilepsy · gamma power · power normalization · seizure onset zone · AUROC · interictal EEG · stereoelectroencephalography · intracranial EEG · spectral power
References
2- usesCommon average reference raises median relative gamma rank agreement in two 60 s SEEG records from each of 36 HUP patients sci:sha256:3f5f9f6ae779730a7bf70359ae9ddc56a5032c2e8abdb97c0cf3c127d6082e74
The study used all 36 archived SEEG spectrum files. Each file had two runs. It also used clinical labels, source metadata, the signal retrieval recipe, and check code. The new test compared absolute gamma with gamma divided by total power.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The analysis used the HUP SEEG dataset from OpenNeuro snapshot 1.1.3; the source record identifies this exact dataset DOI and CC0 license.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesUnder exchangeable outcome labels within therapy, a test did not detect a mean group difference in gamma treatment ranks from two 60-second SEEG records in 36 patients
We used its archived absolute gamma features, all 36 patient spectra, source clinical metadata, and preserved retrieval records. We added a fixed patient outcome test within therapy groups. The new package preserves these inputs and checks their hashes.
- usesFixed gamma treatment overlap gave outcome AUC 0.380 within therapy groups in 36 public SEEG patients
The result on absolute and relative gamma informed the fixed choice of absolute gamma before the new patient score was compared with outcomes. We did not use that contact discrimination result as evidence of an outcome association. The inspected publication record is preserved.
- usesExclusion of 55 to 65 Hz changed gamma onset AUC by at least 0.05 in 14 of 35 SEEG patients when higher power was the onset score
Its comparison of absolute and relative gamma scores informed the new question and fixed higher power score direction. We reused the same 35-patient cohort and onset labels. We tested inclusion and exclusion of 55 to 65 Hz.
- usesBaseline division lowered early gamma onset label AUC in 32 ds004100 SEEG patients with common average reference
Its comparison of gamma power and gamma divided by total power gives a related test of a power denominator. We compare this result in the report. Our denominator is pre-seizure power in the same band, and our state is early seizure.