search

PennLINC / qsiprep

Preprocessing of diffusion MRI
http://qsiprep.readthedocs.io
BSD 3-Clause "New" or "Revised" License
140 stars 58 forks source link

Way to use MNI Pediatric Template? #459

Closed dyhan316 closed 10 months ago

dyhan316 commented 2 years ago

I was wondering if there is a way to use MNI Pediatric template, like how --infant option can be used to use the MNI infant template instead of the adult template.

Is there a way to do this in QSIPREP? (for example, by changing the json files?)

Thank you!

mattcieslak commented 2 years ago

Hi!

I suggest using --dwi-only and doing all your infant-specific anatomical work outside of qsiprep. The T1w-based registration is good for getting gray matter atlases into alignment with the DWIs, but is not great for lining up white matter structures. I'd suggest using DSI Studio or MRtrix3 to register the DWI data directly to a template

dyhan316 commented 2 years ago

Thank you for the response!

Is my understanding correct that you are suggesting:

  1. Run freesurfer on the T1w image using MNI Pediatric template (not using QSIPREP)
  2. Run preprocessing only dwi by using dwi-only using QSIPREP
  3. Use the results of the previous step2 and register the dwi image to the preprocessed results of the step1. (Using DSI Studio or MRtrix3)
  4. Use the "registered" dwi image (and the freesurfer outcomes from step1) as the inputs for qsiprep recon-only to do tractography and get .tck files and so on.

(Sorry for asking, as I am not adept at MRI preprocessing, so wanted to make sure if my understanding was right!)

scchueler commented 2 years ago

I want to add to this: I'm using the --infant flag for preprocessing WITHOUT the --dwi-only flag (due to this issue). However, it seems that QSIprep doesn't recognize the flag properly as the transforms in the sub/anat directory are named "...from-MNI152NLin2009cAsym_to-T1w..."

dyhan316 commented 2 years ago

Thank you for your comment @scchueler ! do you mean that you preprocess T1w separately (perhaps using infant recon all), then provide them as inputs to QSIPREP for dwi preprocessing while doing --infant and --dwi-only?

dyhan316 commented 2 years ago

If possible, could you share the command that you used to do this? I am trying to do something like what you said because some of our subjects are 0~2 years old so conventional skull stripping fails, but I can't get the QSIPREP to take in a pre-done freesurfer result as input, especially during qsiprep reconstruction

mattcieslak commented 10 months ago

You should have no trouble skull stripping in recent versions. We've switched to synthstrip