Dear Alexandre,
Yes, raw fetal MRI could be indeed represented using usual MRI BIDS keywords, which is indeed the format adopted by the mialsupersolutiontoolkit (MIALSRTK).
You can have a basic dataset example illustrated here.
In this example, only the entity “run-” is used to identify the different scans.
If you wish to encode the type of sequence used in the filename, you could use for instance:
sub-01/anat/
sub-01_scans.tsv
sub-01_scans.json
sub-01_acq-HASTE_run-1_T2w.nii.gz
sub-01_acq-HASTE_run-1_T2w.json
sub-01_acq-HASTE_run-2_T2w.nii.gz
sub-01_acq-HASTE_run-2_T2w.json
sub-01_acq-HASTE_run-3_T2w.nii.gz
sub-01_acq-HASTE_run-3_T2w.json
sub-01_acq-TrueFisp_run-1_T2w.nii.gz
sub-01_acq-TrueFisp_run-1_T2w.json
sub-01_acq-TrueFisp_run-2_T2w.nii.gz
sub-01_acq-TrueFisp_run-2_T2w.json
sub-01_acq-TrueFisp_run-3_T2w.nii.gz
sub-01_acq-TrueFisp_run-3_T2w.json
where a column “orientation” that specifies the orientation for each (acq-<label>
, run-<label>
) pair is used the scans
file (See here).
The scans.tsv should look like:
acquisition run orientation
HASTE 1 axial
HASTE 2 sagittal
HASTE 3 coronal
By looking at the latest BIDS specs, it states for the run
entity (See here) that:
If different entities apply, such as a different session indicated by ses-<label>
, or different acquisition parameters indicated by acq-<label>
, then run
is not needed to distinguish the scans and MAY be omitted.
In this case, the following would be more adequate in the context of fetal MRI:
[…], then run
MAY NOT be needed to distinguish the scans and MAY be omitted.
In the current version of BIDS specs, it is not possible to use the _desc-<label>_
entity to encode the acquisition direction, as this entity is intended ONLY FOR DERIVATIVES data (See here)
I am not sure to fully understand what you mean by
Also different in-line (console) processing can be applied (e.g, motion correction, distortion correction).
Are these operations performed during acquisition at the scanner level, or as a post-acquisition processing steps, such as in MIALSRTK to perform super-resolution reconstruction?
In the second case, data is considered as derivative and not raw, and should follow as close as possible the BIDS derivatives extension for MRI data. For now, the documentation of MIALSRTK outputs can give you an example of the organization and naming of the produced outputs (See here).
Hope this helps!
Best,
Sebastien