Add support for input data in BIDS format

[p-j-smith]

Motivation

Suggested by Chris via email:

Most the of the imaging data (of ENIGMA-PD and otherwise) is nowadays organized according to BIDS format. I would advise to either modify the script so that it can directly work on data formatted according to BIDS (and save the output to the so called derivatives folder) or if that is too much work to add a script to the container (that you for example can call with apptainer exec Enigma-PD-WML.sif bids-2-input.sh) that converts/transfers the data from the bids folder to the input folder you need to run the WML pipeline.

Pitch

• [ ] modify the analysis script to handle data formatted according to BIDS
• [ ] save the output to the derivatives folder

Alternatives

No response

Additional context

No response

[K-Meech]

BIDS is predominantly a folder structure and naming scheme for organising nifti files from multiple subjects and sessions. While our current data structure is (for subject ids 1 and 2):

data
    ├───1
    │      └───niftis
    │               ├───T1.nii.gz
    │               └───FLAIR.nii.gz
    │
    ├───2
    │     └───niftis
    │              ├───T1.nii.gz
    │              └───FLAIR.nii.gz
    └───subjects.txt

A BIDS one might be like:

data
    ├───sub-1
    │      └───anat
    │               ├───sub-1_T1w.json
    │               ├───sub-1_T1w.nii.gz
    │               ├───sub-1_FLAIR.json
    │               └───sub-1_FLAIR.nii.gz
    │
    └───sub-2
            └───anat
                     ├───sub-1_T1w.json
                     ├───sub-1_T1w.nii.gz
                     ├───sub-1_FLAIR.json
                     └───sub-1_FLAIR.nii.gz

BIDS may also optionally contain another level of directories inside the subject folder for individual scanning sessions.

So main changes are:

• subject folders must have the prefix sub- before the id
• niftis must be renamed to anat
• nifti filenames must be in the form sub-ID_T1w.nii.gz (for T1) and sub-ID_FLAIR.nii.gz (for FLAIR). There may also be a session id here, if there was more than one scanning session e.g. sub-ID_ses-1_T1w.nii.gz
• Ideally each nifti file has a corresponding .json file containing image metadata/parameters of the scan. This isn't used by the pipeline, but is part of the BIDS format
• Output should be written to the derivatives folder inside BIDS
• We would need to handle providing an optional session id to tell docker which session to process (if more than one exist)
• We could get rid of the subjects.txt file entirely and just pull subject IDs directly from the sub- folder names in the BIDS data

[K-Meech]

Some useful links:

• BIDS starter kit: really clear introduction to the BIDS format, starting from the folders page

• Examples of BIDS files

• BIDS validator: can be used to check if datasets comply with BIDS

• BIDS Derivatives folder information

[K-Meech]

Note that the MRIcroGL dicom to nifti converter will produce BIDS jsons (also called 'BIDS Sidecar') automatically. There's an option on the left side for 'Create BIDS Sidecar', which should be set to 'Yes, Anonymized' by default. If you give the output filename as sub-%i_%p it'll get you pretty close to the filename you need (presumably without the T1w / FLAIR endings?)