BIDS acq_time is unavailable for 216 of 279 OpenNeuro SEEG subjects, and only two datasets pass multi-run checks

We assessed whether BIDS scans.tsv acq_time can provide recording clock times for time-of-day (circadian) SEEG studies. We screened up to eight subjects in each of 42 OpenNeuro datasets (279 subjects), using dataset snapshot versions recorded in the results. In the fixed subject classifications, 98 subjects had no scans.tsv, 118 had no acq_time values, 36 had one timed run, 13 failed a run-overlap check, and 14 passed it. Thus, acq_time was unavailable for 216 subjects. In ds004100, all eight tested subjects had times seconds apart on 2021-08-22 or 2022-08-28, consistent with BIDS conversion stamps rather than recording times; ds004194 failed the overlap check in five of eight subjects. ds004993 had only 00:00:00 times. Only ds003688 (8/8 subjects) and ds003848 (3/3) had multiple timed runs that passed the check; both had shifted dates (1899–1901) but plausible clock times. Five other candidate datasets had only one timed run per subject, which cannot be checked for overlap. A pass is necessary but not sufficient because a constant clock shift is not visible in the files. The analysis tested at most eight subjects per dataset. The recorded scan and run metadata came from the OpenNeuro S3 mirror, which can change across snapshots; the source report does not establish byte-for-byte identity with an immutable DOI snapshot. The results therefore support further verification of ds003688 or ds003848, or obtaining recording times from curators, and show that ds004100 acq_time should not be used as a clock time.

stereoelectroencephalography · BIDS · acq_time · OpenNeuro · circadian studies · data quality · recording time