Hi everyone,
I’m having issues with dwifslpreproc in subjects where I know it should run correctly because I have previously successfully processed them with micapipe which also uses dwifslpreproc. The processing for one subject remains unfinished even after 12 hours, but no error message is shown.
This is a multi-shell dataset with 104 directions in total and 3 shells, I have AP and PA data so 208 volumes in total.
I’m running the following commands using MRtrix 3.0.8 and Rocky Linux 8.10:
$ mrconvert AP_dwi.nii.gz AP_dwi.mif -fslgrad AP_dwi.bvec AP_dwi.bval
$ dwidenoise AP_dwi.mif AP_dwi_denoised.mif
$ mrdegibbs AP_dwi_denoised.mif AP_dwi_denoised_unr.mif
$ # mrconvert, dwidenoise and mrdegibbs for PA images
$ mrcat AP_dwi_denoised_unr.mif PA_dwi_denoised_unr.mif AP_PA.mif -axis 3
$ dwifslpreproc AP_PA.mif AP_PA_preproc.mif -rpe_all -pe_dir ap -readout_time 0.0708501 -eddy_options " --data_is_shelled --verbose --slm=linear"
I tried running this without --data_is_shelled and without --slm=linear, both with the same outcome.
The last few lines of the log file when running the last step are:
Command: mrconvert dwi.mif -import_pe_table dwi_manual_pe_scheme.txt eddy_in.nii -strides -1,+2,+3,+4 -export_grad_fsl bvecs bvals -export_pe_eddy eddy_config.txt eddy_indices.txt
Command: eddy_cuda11.0 --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --data_is_shelled --verbose --slm=linear --out=dwi_post_eddy --verbose
dwifslpreproc: CUDA version of 'eddy' was not successful; attempting OpenMP version
Command: eddy_cpu --imain=eddy_in.nii --mask=eddy_mask.nii --acqp=eddy_config.txt --index=eddy_indices.txt --bvecs=bvecs --bvals=bvals --topup=field --data_is_shelled --verbose --slm=linear --out=dwi_post_eddy --verbose
dwifslpreproc runs successfully on the same system on data that doesn’t have two PE directions where I instead used Synb0 and -rpe_none. Does anyone have an idea what might be going on? Thank you in advance!