Output of reconstruction workflow in Qsiprep seems incorrect

Summary of what happened:

I used the dsi_studio_gqi and mrtrix_singleshell_ss3t_ACT-hsvs to test the reconstruction of dwi image. And it seems that both of them got unexpected results compared with examples from Andy’s Brain Book (I’m so sorry that I’m new to dwi and can only check the results in this unscientific way )

Command used (and if a helper script was used, a link to the helper script or the command generated):

I first run fmriprep:

> export SINGULARITYENV_FS_LICENSE=/freesurfer/freesurfer/license.txt
> singularity run --cleanenv fmriprep/fmriprep-23.1.3.simg \
>    /Data/ds00xxxx-download /Data/ds00xxxx-processed \
>     participant \
>     --participant-label dmp0188 \
>     -w  /Data/ds00xxxx-processed/work \
>     --write-graph

Then Qsiprep:

singularity run --cleanenv /Code/simg/qsiprep/qsiprep.sif \
   /Code/Data/ds00xxxx-download /Code/Data/ds00xxxx-recon-act_ \
    participant \
    --participant-label dmp0188 \
    --fs-license-file /freesurfer/freesurfer/license.txt \
    -w  /Code/Data/ds00xxxx-processed/work/test \
    --write-graph \
    --nthreads 64 \
    --omp-nthreads 32 \
    --freesurfer-input /Code/Data/ds00xxxx-processed/sourcedata/freesurfer/  \
    -vv \
    --recon_spec mrtrix_singleshell_ss3t_ACT-hsvs\
    --recon_input /Code/Data/ds00xxxx-processed/qsiprep/

Environment (Docker, Singularity, custom installation):


Screenshots / relevant information:

This the output of reconstruction:

Examples from Andy’s Brain Book:

Please forgive me if I ask very basic questions.


Hi @taka2223,

What version of qsiprep did you use?

Also, I believe what you’re looking at is the ODF plot, not the streamlines. The best way to check for tractography quality would be to load streamlines into some viewer like mrview and see if the terminations happen right at the gray matter white matter interface, which would mean anatomically-constrained tractography did its job.



Thank you very much for your response!

The version of qsiprep I used is qsiprep v0.19.0

Yes indeed. I’m sorry that I confused these two concepts due to my poor knowledge. :dizzy_face: :dizzy_face:

Thanks for your suggestions and actually I have tried mrview after posting the question. But unfortunatly another problem is that I use MRtrix on HPC and I got stuck in running mrview as it should run on a GUI support system. And I also failed to use MobaXterm to open the viewer in my Windows system. Would you mind recommending some other visualization tools or provide any suggestions you may have regarding this issue? I would greatly appreciate your insight. :blush: :blush:

It seems that the most straight forward method is using the MRtrix on windows and transmitting the data from remote server?

That would be a good idea, or you can make visualizations with DIPY, although you’d have to write code for that. See DIPY : Docs 1.7.0 - Examples

1 Like

On our system I always copy (via rsync, scp or datalad) the sh mif files to my local machine and view it in a local copy of mrview. I know some people have been able to view ODF data on remote machines with VNC viewer but I prefer the responsiveness of running a local copy of mrview

1 Like