MRI Data Injection¶

Data converted with the ANC conversion pipeline is either automatically converted from the scanner or can be historical data.

For historical data, start with converting a single subject/session. If there are multiple sessions where different data is acquired, convert and merge one instance of each session into main before other data is converted.

For newly acquired data, pilot data or first subject/session data should be prioritised by data stewards. Additional subject/session merge requests may be added before the first data is merged in — handle these cases with care.

The following workflow applies for each MRI data injection MR.

After conversion¶

• MR is assigned to a user
• User has followed the handbook
• When everything is correct, they mark the MR as ready
• Data steward is assigned as reviewer

Review checklist¶

• Check the data overview (assume "ready" means all expected data is there, but check for uncommented inconsistencies)
• Check filenames (conversion and validation should ensure correctness, but check for inconsistencies)
• Check events (ping Monique for first subject/session)
• Check participants (currently not caught by the BIDS validator)
• Is the number of files correct for all participants? (if data is missing or excessive for a subject, the user should create an issue to add this to the README)
• Is all participant information in the tsv (participant_id, sex, age, diagnosis)?
• Is all information present in the json files (participants + events)?
• Are all functional sequences belonging to a fieldmap listed in that fieldmap's .json file?
• Do all files (especially functional) have the expected size?
• If DWI data was acquired: was the correct direction label assigned and is the phase encoding direction correctly labelled in the corresponding .json?
• If multiple sessions were acquired: is there a sessions.tsv?