Across 35 ds005624 sessions from 24 patients, no event markers were found; DC1 pulses appeared in 26 sessions, with amplitudes about 1e6 too large

We audited all 35 sessions of OpenNeuro ds005624 snapshot 1.0.0 (24 HUP patients, BIDS task label CCDT). All 35 sessions have 0 markers in the vmrk files, no events.tsv file, and TriggerChannelCount 0. All 5002 channel rows have type SEEG and status good. But every session has channels that are not SEEG: DC1 (30 sessions), DC7 (17), EKG (6), LOC/ROC (13), and scalp EEG names (14). We read two 4-s windows from each session with HTTP byte ranges. In 26 sessions of 20 patients, DC1 has the same pulse type: width 9.8-10.7 ms, peak/noise approximately 14000, and onset intervals 0.165-2.8 s. These pulses are probably task event codes. The meaning of the code is not documented. In two 60-s windows of HUP191 ses-1, the pulses occur in groups of 1-4, with in-group intervals of 0.166-0.186 s or 0.320-0.342 s. DC1 does not have usable pulses in 9 sessions: no DC channel (HUP133 x4, HUP136), a 60 Hz signal (HUP184 x2, full session checked), all zeros (HUP187 ses-1), and weak 2-ms transients only (HUP171). DC7 has no pulses. The SEEG SD, scaled by the header resolution, is 123-464 V. Thus the stored values are approximately 1e6 times too large. We assume that microvolt values were written as volts. In HUP178 ses-1, the data length (1138.75 s) does not agree with RecordingDuration (847.7 s). Limits: most sessions were sampled only in 8 s. The package has the code, the byte-range probe data, the SHA-256 provenance and the reproduction steps.

stereo-EEG · BIDS · event timing · channel metadata · amplitude scaling · OpenNeuro ds005624