Edges of the topup_fieldcoef image have abnormal signal

[treanus]

Hi,

thanks for making synb0 available.

Recently I have a number of datasets where the edges of the fieldcoef image look strange. The top few rows and the anterior few rows contain wrapped signal. See:

This leads to abnormal signal in the output of eddy. It as as if the signal from the eyes is aliased a few times at the far left of the processed dwi images.

Is this related to the FOV of the T1 and dwi not being equal?

[iPsych]

@treanus I am experiencing same issue. I guess that, possibly it's due to wrong acqparams.txt value. Any advance in this issue?

[schillkg]

That is indeed strange. We can do a quick debug to figure out whether it is the synthesis or a topup-related issue. Does the synthesized b0 image look reasonable (b0_u.nii.gz, or the 2nd volume in b0_all.nii.gz) ? It should have the same contrast as the distorted b0, but should look noticeably less distorted (and smoothed just a little bit). If this looks good, it is probably a topup issue - I'm not sure that is wrapped signal? I don't think we expect anything in that area from the posterior of the brain? it just seems like topup isn't well estimating the field near the eyes and is doing something funny with that signal?

[robbwh]

potentially related: https://github.com/MASILab/SynBOLD-DisCo/issues/1

[iPsych]

@robbwh @schillkg It seems the possible topup issue. The first picture is b0_all_topup.nii.gz, which looks not that problematic, but the 3nd and 3rd pictures, the results of fsldwipreproc with topup options using below command gives distorted output, especially first and last slides.

wi_den_unr_preproc.mif -pe_dir AP -rpe_none -eddy_options " --slm=none --data_is_shelled" -fslgrad INPUTS/bvec INPUTS/bval -topup_files OUTPUTs/topup_