In 278 clear SEEG recordings, measured mains frequency never disagreed with a numeric BIDS value; 131 of 409 recordings were unclear

This descriptive audit asked whether BIDS PowerLineFrequency agrees with the mains frequency measured in public OpenNeuro stereo-electroencephalography (SEEG) signals, and whether BIDS SamplingFrequency agrees with the data-file header. From an archived SEEG census, the first five subjects per dataset were selected, except that all 254 SEEG subjects in ds004789 were included. One EDF or BrainVision file and one 10 s window per subject were examined using line scores at 50 and 60 Hz and their harmonics; classification thresholds were fixed in a plan written before signal data were read. Of 435 attempted subject jobs, 409 subjects in 35 datasets had readable files. Among 278 recordings with a clear measured line and numeric sidecar value, 264 agreed and none disagreed (12 measured 50 Hz and 252 measured 60 Hz). The other 131 of 409 recordings (32%) had no clear peak. Five tested datasets provided no numeric PowerLineFrequency value: four used “n/a” and one had no tested-file sidecar; 13 of 15 tested subjects in three of those datasets had a clear 60 Hz line. SamplingFrequency agreed with the header in 408 of 409 subjects; the exception had no sidecar. This audit used one file and one window per subject, sampled only five subjects in each dataset except ds004789, and cannot detect a wrong mains value when the line is unclear, including when a notch filter may have removed it. The thresholds are fixed descriptive choices, not calibrated against ground truth, and sampling-rate agreement does not validate the true clock. Signal files were read from recorded OpenNeuro S3 URLs on 2026-10-11/12; exact dataset release applicability is not established. The package preserves the preread plan and hash, code, file keys, and subject-level results.

stereo-electroencephalography · SEEG · BIDS · PowerLineFrequency · mains frequency · line noise · OpenNeuro · sampling frequency · data audit