Dwifslpreproc - Eddy failed with message

Dear MRtrix community,

Although another topic about this error was poster previously (Dwifslpreproc Fails with EDDY::: Eddy failed with message Error), I did not find a solution yet.

I’m trying to run dwifslpreproc on my multi-shell neonatal data, including a B0 pair, using the following command:

for_each * : dwifslpreproc IN/dwi_denoised_unringed.mif IN/dwi_denoised_unringed_preproc.mif -pe_dir j- -rpe_pair -se_epi IN/B0_pair.mif 

However, I receive the following error:

dwifslpreproc: [ERROR] eddy_cpu --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --out=dwi_post_eddy --verbose (dwifslpreproc:972)
          dwifslpreproc: [ERROR] Information from failed command:
                         Reading images
                         EDDY::: Eddy failed with message 

Once I open the textile ‘error.txt’ in the scratch directory, I see no additional error information.
The images loaded for the eddy_cpu command can be opened manually using mrview and are present in the scratch directory.

Any advice on how to continue?

Many thanks in advance!!


@ejanson6 Im the original poster of the question you referenced. I think this may be an issue with mac OS and fsl eddy/cuda compatibility but I am not certain.

Hm it looks like the following command failed:

eddy_cpu --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --out=dwi_post_eddy --verbose (dwifslpreproc:972

did you check in the temp directory if the input imgages (eddy_in.nii, eddy_mask.nii, eddy_config.txt ) exist?
Do they look fine or are they faulty?

+1 to this - I’ve seen this eddy error before and in my case, the auto-generated eddy_mask.nii file was way off from a proper brain mask. I was able to circumvent this by using the -eddy_mask argument for dwifslpreproc since I already had a mask from some other processing step.

Thanks for your response! In my case, the eddy_in and eddy_mask file look okay to me. See here the mask with reduced opacity on top of the input file:

Dear mschira, thank you so much for thinking along. These files do exist, can be opened and look good.

Do you have fsl installed? If so I would recommend trying the command:
eddy_openmp --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --out=dwi_post_eddy --verbose

That should do exactly the same as eddy_cpu, but forgoing the mrtrix wrapper.

Dear mschira, thank you for the suggestion. FSL is installed, but the command eddy_openmp and the ‘normal’ eddy do not work. I still receive the error message about reading images…

dwifslpreproc-tmp-1WCNWH % eddy_openmp --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --out=dwi_post_eddy --verbose

zsh: command not found: eddy_openmp

dwifslpreproc-tmp-1WCNWH % eddy --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --out=dwi_post_eddy --verbose

Reading images


Maybe that is where the problem lies?
I think eddy_cpu calls essentially eddy_openmp.
Can you try and locate if the program eddy_openmp exists on your computer maybe using the search function? What system are you using?
I don’t think troubleshooting your FSL or MRtrix installation is my forte.
I would recommend you to use the Neurodesk app. You can find it at neurodesk.org.
We have working modules for MRtrix 3.0.2 and 3.0.3.

I would probably recommend in the neurodesk app you would need to type

ml mrtrix3/3.0.3

and then do your dwifslpreproc again.

If you are running it on a server you can use neurocommand. Let me know, I can help you.