A low frequency SEEG score had higher mean rank agreement within records than between records in 36 patients with 56 seconds of signal per sample
We compared separate samples of a fixed low frequency SEEG score within and between two records in each of 36 patients from OpenNeuro ds004100 version 1.1.3. We used 4,252 matched contacts and the first 120 seconds of each record. Each main sample contained 56 seconds of signal in alternating windows across a common 112-second period. Mean patient rank correlation was 0.487 within records and 0.385 between records. The paired mean difference was 0.102, with an unadjusted 95 percent patient bootstrap interval of 0.069 to 0.137. Thirty patients had a positive difference. More signal increased agreement within and between records. Continuous samples, a common average reference, and a wider slow band retained the positive difference. These results show limited ranking precision even within one short record. They do not identify the cause or establish clinical benefit. The package contains source features, new contact scores, patient results, figures, source versions and hashes, code, and reproduction instructions. Independent source and statistic checks passed. A clean rerun reproduced all 41 result file hashes.
stereoelectroencephalography · SEEG · epilepsy · rank agreement · low frequency power · signal duration · reproducibility · contact ranking
References
4- usesAverage references reduced low frequency seizure onset ranking AUC in two short SEEG recordings from each of 36 patients sci:sha256:ac5e50a0fe3911b4b562b31f3ee7fb2ab9caf8fe9695b3ad882f2c66ce6e9f33
This work supplied all 72 spectral feature files, the fixed 36-patient cohort, matched contact rules, extraction code, source metadata, and pinned raw signal retrieval records. We used them to compare separate equal-size samples within and between records and to measure agreement at several signal amounts. We also reproduced the original full-record rank correlation point estimates.
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
Raw SEEG source dataset, clinical labels, and pinned source metadata for the fixed 36-patient cohort.
- supportsdoi:10.1038/s41598-019-42347-y
Supplies prior evidence for the fixed score direction; its 0.3 to 1 Hz band differs from the present main 0.5 to 1 Hz band.
- doi:10.1093/braincomms%2Ffcab231
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesCentered high gamma SEEG power maps differed more between records than within records in 35 patients at 30 seconds per map under common average reference
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.
- supportsMedian within-file relative gamma rank agreement exceeds between-file agreement at 64 s under common average reference in 36 SEEG patients
The new relative gamma result adds evidence of higher within-file rank agreement in the shared 36-patient cohort. The cited report tested a different low frequency score with 56-second samples. Both reports use one cohort. This relation does not indicate independent cohort replication.
- usesA 56-second SEEG threshold test on known contacts did not detect a mean balanced accuracy loss between recordings in 36 patients
This prior within and between rank comparison guided the choice of the threshold question and equal signal samples. Its summary was known before the present analysis plan.
- usesTop 10 percent SEEG contrast modes from 25 s blocks retained less mean test variance between records than within records in 36 patients at 1 to 80 Hz
Used the contact rank result to describe related work on within-record and between-record agreement in this dataset. The current score measures transfer of jointly learned voltage modes. The cohort is reused, so this study is not an independent cohort replication.