display warp: fullWarp_abs.nii.gz is invalid: should be encoded in a 5D file with vector intent code

[sandrinebedard]

Description

I wanted to apply sct_image -i fullWarp_abs.nii.gz -display-warp to the warping field (fullWarp_abs.nii.gz) used for correction of gradient nonlinearities (gradient_unwarp.py). I got this error:

(venv_sct) sabeda@DESKTOP-JQ8A4UV:/mnt/c/Users/sb199/Projet3_data$ sct_image -i warping_field/fullWarp_abs.nii.gz -display-warp

--
Spinal Cord Toolbox (git-jca/3336-display-warp-7de29b49e9979eb22409e8a11eb2c79eb0277281)
sct_image -i warping_field/fullWarp_abs.nii.gz -display-warp
--

Creating temporary folder (/tmp/sct-20210407144012.007017-jex9nz5p)
sct_resample -i grid_3.nii.gz -f 3x3x1 -x nn -o grid_3_resample.nii.gz # in /tmp/sct-20210407144012.007017-jex9nz5p
sct_apply_transfo -i /tmp/sct-20210407144012.007017-jex9nz5p/grid_3_resample.nii.gz -d /tmp/sct-20210407144012.007017-jex9nz5p/grid_3_resample.nii.gz -w /mnt/c/Users/sb199/Projet3_data/warping_field/fullWarp_abs.nii.gz -o /mnt/c/Users/sb199/Projet3_data/warping_field/grid_3_resample_fullWarp_abs.nii.gz # in /mnt/c/Users/sb199/Projet3_data
Traceback (most recent call last):
  File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_image.py", line 554, in <module>
    main(sys.argv[1:])
  File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_image.py", line 233, in main
    visualize_warp(im_warp=im_in, im_grid=None, step=3, rm_tmp=True)
  File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_image.py", line 547, in visualize_warp
    run_proc(['sct_apply_transfo', '-i', fname_grid, '-d', fname_grid, '-w', fname_warp, '-o', grid_warped])
  File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/utils/sys.py", line 320, in run_proc
    raise RuntimeError(output)
RuntimeError:
--
Spinal Cord Toolbox (git-jca/3336-display-warp-7de29b49e9979eb22409e8a11eb2c79eb0277281)

sct_apply_transfo -i /tmp/sct-20210407144012.007017-jex9nz5p/grid_3_resample.nii.gz -d /tmp/sct-20210407144012.007017-jex9nz5p/grid_3_resample.nii.gz -w /mnt/c/Users/sb199/Projet3_data/warping_field/fullWarp_abs.nii.gz -o /mnt/c/Users/sb199/Projet3_data/warping_field/grid_3_resample_fullWarp_abs.nii.gz
--


Parse list of warping fields...
Traceback (most recent call last):
File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_apply_transfo.py", line 365, in <module>
main(sys.argv[1:])
File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_apply_transfo.py", line 360, in main
transform.apply()
File "/mnt/c/Users/sb199/spinalcordtoolbox/spinalcordtoolbox/scripts/sct_apply_transfo.py", line 169, in apply
.format(path_warp))
ValueError: Displacement field in /mnt/c/Users/sb199/Projet3_data/warping_field/fullWarp_abs.nii.gz is invalid: should be encoded in a 5D file with vector intent code (see https://nifti.nimh.nih.gov/pub/dist/src/niftilib/nifti1.h

I tested sct_image -display-warp with a warping field created with sct_register_multimodal and everything worked.
So I think fullWarp_abs.nii.gz follows FSL convention and not ITK, it has only 4 dimensions.

Maybe not worth the trouble for just displaying the grid with the warping field.

The file is in : duke/temp/sebeda/data_warping_field/fullWarp_abs.nii.gz

[sandrinebedard]

UPDATE

I did not figure out how to convert the warping field in the ITK format to make it compatible with sct_image -display-warp but, I figured out that fullWarp_abs.nii.gz is a volume of coefficient, so to have the field, I ran :

fnirtfileutils -i fullWarp_abs.nii.gz -o field.nii.gz

And on ITK-SNAP, when clicking on multi-component display --> grid I get this:

Well, it is not as nice as the output of sct_image -display-warp but, from what I observed, the grid is only in the axial view, I think having the sagittal view for the warping field may be better here, maybe we ca do something with that image instead.

[sandrinebedard]

@jcohenadad I may need your insights on this please 😊

[jcohenadad]

So I think fullWarp_abs.nii.gz follows FSL convention and not ITK, it has only 4 dimensions.

yup! you figured this out. As you probably noticed we have a sct_image -to-fsl feature, but no sct_image -to-itk 😬

However, your investigation in #59 (comment) is excellent, and i think largely sufficient if the goal is simply to export a nice-looking image to illustrate gradient non-linearity correction in the pipeline figure of your article

[sandrinebedard]

However, your investigation in #59 (comment) is excellent, and i think largely sufficient if the goal is simply to export a nice-looking image to illustrate gradient non-linearity correction in the pipeline figure of your article

That's what I thought! thanks