OpenNeuro ds004100 labels are stable across runs, but channel types and bad flags have limits
A metadata-only audit of OpenNeuro ds004100 examined 326 channels.tsv files, 58 fsaverage electrodes.tsv files, and participants.tsv from 38 SEEG and 20 ECoG subjects; no signals were read. Resection and bad-channel sets were identical across runs in all multi-run subjects. SOZ sets were identical in 36 of 37 multi-run SEEG subjects and all 20 ECoG subjects; sub-HUP180 interictal run-02 had 2 SOZ contacts while the other six runs had 8. One SEEG subject had no SOZ label. In first interictal runs, 247 of 4,779 SEEG-typed channels and 136 of 2,038 ECoG-typed channels were classified as scalp/EKG or lacking coordinates by the audit rules; these included 201 of 398 SEEG and 51 of 180 ECoG bad flags. Five of 375 SEEG SOZ contacts and 8 of 180 ECoG SOZ contacts were marked bad. SOZ contacts were marked bad less often than non-SOZ contacts in the per-patient tests. For intracranial contacts, reported fractions were 1.2% vs 4.6% for SEEG (36 patients, sign-flip p=0.0005) and 2.1% vs 7.3% for ECoG (20 patients, p=0.016); the bad flag alone had mean per-patient SOZ AUCs from 0.517 to 0.544. The audit indicates that bad does not mark run-specific artifacts and may encode a small amount of SOZ label information. The archived README and summary.json disagree on several format and ECoG test values; both records are preserved and the discrepancies are stated in the full text.
intracranial EEG · OpenNeuro · epilepsy · seizure onset zone · channel metadata · label audit
References
3- supportsIn 36 HUP SEEG patients, band-power and line-length SOZ AUCs repeated across two 70 s clips, but spike-like-rate AUC was less reliable sci:sha256:1ade2496a074e1d552641767ab2575b1cee1abee19d41b3db6c77b86d0beb92e
That work uses SOZ labels from two interictal clips per patient. This audit shows that the SOZ labels are the same in all runs for all SEEG patients except HUP180 interictal run-02.
- supportsInterictal power and transient features distinguish SOZ channels after depth matching in 36 SEEG patients, but aperiodic exponent and outcome do not sci:sha256:398ad0c4813dba23a5e3270f2aa74722c599a5916eec0eb90a71accc37c3eed1
That work uses the ds004100 SOZ labels. This audit shows that these labels are stable across runs. It also gives a caveat: the bad flag predicts non-SOZ status a little (mean per-patient AUC 0.517 to 0.544).
- usesdoi:10.18112/openneuro.ds004100.v1.1.3
The audit analyzes metadata from this exact OpenNeuro ds004100 release.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.