Does dwigradcheck provide clues on how to change the header info, and how to do that?

1

Here is the result of the dwigradcheck of the merged AP PA image using dwifslpreproc and it’s resulting bvec/bval. PS: The tensor fitting looks bad for this.

As I understood, none 0,1,2 should be first to indicate a match between the file and the bvec?

If so how should I change my bvec table to fit the .nii file?

Mean length Axis flipped Axis permutations Axis basis
33.95 1 (0, 1, 2) image
33.46 1 (0, 1, 2) scanner
15.51 2 (0, 2, 1) scanner
15.46 1 (1, 0, 2) image
15.42 0 (1, 0, 2) scanner
15.36 1 (0, 2, 1) image
15.10 1 (0, 2, 1) scanner
14.95 2 (0, 2, 1) image
14.83 1 (2, 1, 0) scanner
14.81 1 (1, 0, 2) scanner
14.80 none (2, 1, 0) scanner
14.79 1 (2, 1, 0) image
14.75 none (2, 1, 0) image
14.66 0 (1, 0, 2) image
14.52 0 (0, 1, 2) image
14.31 0 (0, 1, 2) scanner
14.11 2 (0, 1, 2) scanner
13.93 2 (0, 1, 2) image
13.47 none (0, 1, 2) scanner
13.45 none (0, 2, 1) image
13.35 0 (0, 2, 1) scanner
13.34 0 (2, 0, 1) scanner
13.34 none (0, 1, 2) image

1 Like
2

Indeed.

If you use dwigradcheck with one of the the -export_grad_* options, you can have it output the corrected gradient directions, e.g.:

dwigradcheck dwi.nii -fslgrad bvecs bvals -export_grad_fsl bvecs_corrected bvals_corrected

(make sure to now use the corrected bvecs and bvals for all subsequent commands that require gradient directions) or much simpler using .mif:

dwigradcheck dwi.mif -export_grad_mrtrix grad.txt
mrconvert dwi.mif -grad grad.txt dwi_corrected.mif
1 Like