What is the proper #BIDS format for pre and post contrast T1w?

Hello all,

I have anatomical scans with both pre- and post-contrast T1w. Is the following format correct? How can I specify whether the scan is 2D or 3D?


For post-processing, I’m registering all anatomical files into the same space. I would appreciate advice on the best practices for naming these registered files and their proper storage location within the BIDS framework.

I also have double inversion recovery (dir) scans, which suffixes should they use?


For 2D or 3D my first hunch would be to rely on the acquisition entity:

For example see below.


The filename templates for anatomical file can be found here:

Note that the acq entity comes before the ce one: the order of all the entities is always the same in BIDS and can be found here:


This is more for derivatives BIDS.

There are some rules that can already help:


For example you could go the way below (would require you to know what space the images are coregistered to).


Thank you Remi! That’s very helpful.

The space the images are coregistered to are subject specific. All anat files are registered to an averaged T1w space across multiple sessions for each subjects. Will the following structure be okay?

└─ derivatives/
   └─ pipeline_xxx/
      └─ sub-001/
         └─ ses-01/
            └─ anat/
               ├─ sub-001_ses-01_acq-2D_space-individual_desc-preproc_FLAIR.nii.gz
               ├─ sub-001_ses-01_acq-2D_space-individual_desc-brain_mask.nii.gz
               └─ sub-001_ses-01_acq-2D_space-individual_desc-lesion_mask.nii.gz
         └─ ses-02/
            └─ anat/
               ├─  sub-001_ses-02_acq-2D_space-individual_desc-preproc_FLAIR.nii.gz
               └─ sub-001_ses-02_acq-3D_space-individual_desc-preproc_FLAIR.nii.gz

Is acq entity included for the post-coregistered files as well? The original 2D scans are now registered to 3D space.

Yes but to make more explicit what “individual” means BIDS requires you to specify to what image things were coregistered to.

See this example: Common data types and metadata - Brain Imaging Data Structure 1.10.0-dev

Yes in general for derivatives naming you keep the “basename” and just append things to it.
The exception is when some processing operation makes an entity “obsolete”: say you have several runs of functional data


And you compute a mean across runs. Then you would drop the run entity when naming the mean image.

1 Like