Dear Corinna,
can you please let me know which exact version of eddy your are using? I.e. what FSL versions/pre-release and if it is the GPU or the openmp version?
> Hello all,
>
> I have 30 direction DKI data (b=1000,2000) from a Phillips machine and I am having a few difficulties with eddy.
>
> This is my sample command line:
>
> eddy --imain=rawdki.nii.gz --mask=b0_brain_mask.nii.gz --acqp=acqparams.txt --index=index.txt --bvecs=bvec.bvec --bvals=bval.bval --niter=5 --repol --out=subject_eddy_corrected_dki_repol
>
> A number of subjects fail when --repol is included. I get the attached error. The subjects (all adults) whose data fails all have some form of brain injury, most with large ventricles due to prenatal hypoxic-ischaemic injury.
>
It looks from the crash dump that there is a problem with the interpolator indexing outside of the image volume. A couple of releases ago there was a problem that might cause that to happen. I _think_ that is no longer the case, but it would be good to know what version you use.
> As a results, I processed everyone without the --repol flag and it runs well, except that I do not get an eddy_rotated_bvecs output file.
>
> 1. any suggestions for how to get repol to work?
Judging from the crash dump it shouldn’t really have anything to do with --repol.
> 2. How to get the rotated becs?
If you have a sufficiently new version these should be written automatically. But they are written after the corrected images are written, so there is a possibility that eddy crashed between writing the corrected images and writing the rotated bvecs. If that is the case you should be able to finding an error message from your queuing system also for those case that appears to have finished.
Jesper
>
> Thank you,
>
> Corinna
> <error.txt><acqparams.txt>
|