Orientation question when applying MNI transform to DTI data/bvec

AFNI Message Board
1

Hi AFNI experts,
I had a question regarding vector orientation information when applying MNI transforms to DTI data . I have a file that is eddy current corrected and its corresponding bval/bvec corrected files in subject space. I have a T2 image that was then aligned with the DTI data. The original brain (not B0 aligned) is also rigidly aligned to the MNI template and then ran auto_warp.py (which does an affine and nonlinear warp). I concatenated rigid, affine, nonlinear into a single transform and the T2 and DTI data warped correctly. I used cat_matvec to concatenate the rigid alignment and affine transforms since 3dAllineate outputs all transforms in DICOM order (RAI) so it is not a problem that the aligned images are in different orientations (as MNI was LPI orientation, T2 was RPI and DTI is ASL orientation) and then I extracted the rotation component. I want to use this transform with vecwarp program to rotate the corrected bvec file so that they are in MNI space and can be used with explore DTI or other programs etc. since the warps to MNI came out great using AFNI.

I wanted to find out if the subject space bvecs (3 columns) is in ASL orientation, can I directly use vecwarp to apply the rotations above to get it aligned with MNI space or do the bvecs have to be converted in to DICOM RAI orientation first since the matrix being applied is in RAI order? Based on the user post below I assume this procedure above is fine, but just wanted to verify since the post was regarding eddy current correction and not specifically rotating to MNI space. I am pretty confident with the cat_matvec part but I wasn’t 100% sure with the use of vecwarp if I need to do anything different for bvec files that are in oblique or other orientations other than RAI.

http://blog.cogneurostats.com/2013/09/06/rotating-bvecs-for-dti-fitting/

Thanks,
Ajay

2

Hi AFNI Experts,
I just wanted to follow up to see if there were any issues using vecwarp with ASL orientation and the 3dAllineate matrix alignment (RAI dicoms orientation) directly to MNI space to correct bvecs for the rotation or if I have to account the initial ASL orientation of the dataset somehow prior to vecwarp.

Thanks,
Ajay

3

The general recommendation around here is to use TORTOISE[/url] for all things diffusion, and you can see the updated instructions on [url=http://blog.cogneurostats.com/2017/04/10/preprocessing-diffusion-weighted-images-with-tortoise-3-part-1/]the blog page you linked to for how to do the preprocessing. If there’s something that TORTOISE doesn’t do, a feature request could be made to add it.

More to your question: Vecwarp was written by Bob, and he’s away at a Bootcamp for the next week.