OpenNeuro ds003029 v1.0.3 has no complete public SEEG seizure recording
This audit asked whether OpenNeuro ds003029 v1.0.3 can serve as a public SEEG dataset with seizure-onset-zone (SOZ) labels. The clinical table contains 100 patients: 65 with SEEG (64 from Cleveland Clinic, whose data are not in the public snapshot) and 35 with ECoG. Of 37 public subjects, the table identifies 33 as ECoG and one as SEEG (sub-umf004); three subjects are not in the table. channels.tsv labels only sub-umf004 as SEEG and contains no SOZ labels. The SOZ is recorded only in the clinical workbook: 412 of 415 expanded SOZ contacts match a channel name, across 33 subjects. Of 105 ictal runs with events.tsv, 80 have both onset and offset markers. Size checks found 102 of 106 ictal EEG files within 1% of the size declared by the BrainVision header and ieeg.json. Four UMF files are truncated: umf002 (0.66%, 1.34 s of 202 s), umf003 (1.0%, 1.97 s of 194 s), umf004 (0.28%, 0.69 s of 248 s), and umf005 (0.46%, 1.19 s of 257 s); umf001 is complete. Thus this snapshot contains no usable public SEEG ictal recording. The interictal SEEG run for umf004 has a channels.tsv but no EEG file in the key listing. File size was checked, not signal content, for 105 of 106 files; a file with the expected size may still contain bad data. The audit covers only snapshot 1.0.3. SOZ matching uses names and does not establish that the clinical and BIDS montages agree. Onset and offset markers are free text, and the rules in audit.py may miss or mislabel markers.
intracranial EEG · SEEG · ECoG · epilepsy · seizure onset zone · OpenNeuro · BIDS · dataset audit
References
2- usesdoi:10.18112/openneuro.ds003029.v1.0.3
The audit examined sidecars and the clinical workbook in the recorded OpenNeuro ds003029 snapshot 1.0.3.
- doi:10.1101/862797
The dataset's source documentation identifies this paper as the work describing the seizure-onset-zone analysis and associated data.
Cited by
1 publicationPublic catalog entries on Sciscape.
- supportsIn OpenNeuro ds003876 v1.0.2, channel metadata classify 30 of 39 subjects as SEEG; only 1 SEEG subject has sleep and wake runs, and 5 ECoG subjects have run-dependent SOZ labels
Both find that umf004 is the only SEEG-typed UMF subject.