participants.tsv multi-session edit

[wulms]

Hi dear BIDS team, wouldn't it make sense to change the participants.tsv to standardize for multiple sessions:

• add a column for session-id
• add a column for AcquisitionDate and PatientBirthDate, PatientSex and PatientWeight from the dicom header.
• calculate age from that information

Best, Niklas

[emdupre]

Hi @wulms --

Thanks for thinking about this ! Currently, the BIDS specification recommends adding an additional _sessions.tsv for each longitudinally collected subject, see here.

Would this address your concerns ? It'd be great to make clearer in the Starter Kit, so please let us know if you have any suggestions to do so !

[wulms]

Hi @emdupre ,

I am thinking about this, because I am applying the BIDS specification with a tool that I am developing in R and implemented the automated writing of the participants.tsv to the sourcedata folder.

Your link informed me, about the current specification of the _sessions.tsv. But, I think age as construct of time difference between AcquisitionDate and PatientBirthDate could be misleading in our study situation. I think it is to some point additional work, to write this file in every participant folder after extraction from each JSON - and to get the information extracted again indexing and reading in every file. Then one can rely simply on the JSON file?

Otherwise, one simply read the participants.tsv containing all the needed information of the cohort, surveys and sensitive information with the additional columns. On the one hand, to get an overview of age distribution, on the other to simplify quality control on the subject-ids.

I would also be happy, to contribute this little thoughts (if seen relevant) and the tool to the BIDS specification.

ID Quality control: I extracted the metadata from all JSONs and compared filename-subject-id with the two header-subject-ids, and can check for incongruent naming, also on implausibilities based on sex, gender or acquisitiondate and birthdate when comparing these variables between the surveys.

Our data situation: We have about 4500 participant's dicom folders from 4 surveys (more or less 2 years apart from each other) and about 100 different naming schemes based on name changes in the scanner. It was to some point quite difficult for me to apply the BIDS specification. But now it works and is reproducible, because all settings a user has to modify are stored in 3 csv-files (session, subject and study). I want to keep the user-interaction as minimal as possible and reproducible. As well as extracting information as easy as possible.

If my tool is ready to test (it runs on our testing branch, but needs more code documentation). Am I allowed to make a push-request for adding it to the existing ones. And is there some way, to validate it externally?