I’m trying to register FLAIR images in subject space to a template space dMRI (b0) image and was wondering about the best way of doing so with FSL or MRtrix. The ultimate goal is to transform a mask that’s in FLAIR space to the template b0 space. Am I correct to assume that FSL FLIRT will likely not work well between subjects, and that FSL FNIRT can only be used within the same modality? My current plan is the following:
I was wondering whether people had other recommendations? I’ve used SPM’s DARTEL before but it seems like an overkill solution here, and I’ve never used MRtrix tools for registrations like these so not sure if that would be a good alternative. I could also imagine modifying my plan to go from FLAIR subject space to FLAIR template space to b0 template space but am not sure if that’s better.
Thank you in advance!
What images do you have in template space? I would imagine the easiest solution would be to simply calculate a registration between flair_subject --> flair_template and then apply that registration to the mask with nearest neighbor interpolation. Since this template space may require non linear deformation, FNIRT (or other nonlinear registration tools) may be the best bet.
in template space I have a b0 image and a WM FOD. I could create a FLAIR image in template space but don’t have it to begin with, which is why in the original plan I wanted to go via the b0 images in subject space.
Sounds like this might be a group B0 template for a fixel based analysis? In which case you probably have the subject_b0 --> template_b0 saved somewhere already, or that is easily calculable. Might help to know all anatomical images you have available. Are your subject spaces in DWI not aligned to anatomical images (such that the subject_b0 --> template_b0 would be accurate for them as well)?
Yes that’s right, it is part of an FBA and I do have the subject2template_warp.mif files, however the issue is that I’d still need to register the FLAIR_subject to the B0_subject with a rigid transformation, and then I don’t see a way how I can concatenate that transformation matrix with the subject2template_warp.mif deformation field.
The other anatomical images I have available are T1w and T2w, none of them are aligned with the diffusion images though. Looks like the FLIRT/FNIRT solution is the one! I just wanted to make sure I’m not missing anything that could be better