EPI to T1w misalignment

[kiragy02]

Summary of what happened:

Dear all,

We’ve encountered a problem of misalignment between EPI and T1W. The Freesurfer bbregister failed to align EPI and T1w as well as Freesurfer mri_coreg. While the recon output seems to be alright in the fmriprep output report but it’s in a different orientation when being viewed in fsleyes. Can you kindly provide any suggestions?

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

/opt/conda/envs/fmriprep/bin/fmriprep /output_fmriprep_new --work-dir /work_fmriprep_new participant --participant_label sub-0001 --bold2anat-dof 12 --dummy-scans 10 --skull-strip-t1w skip --fs-subjects-dir /external_recon/Output --force bbr --output-spaces T1w MNI152NLin6Asym:res-2 MNI152NLin2009cAsym --write-graph --omp-nthreads 4 --nthreads 4 --fs-license-file /license.txt --stop-on-first-crash --mem_mb 40000

Version:

freesurfer/7.3.2
fmriprep/25.0.0

Environment (Docker, Singularity / Apptainer, custom installation):

We run the command in Slurm with a bash script.

Screenshots / relevant information:

---

Many thanks
Yan

[kiragy02]

FYI here is the output of external freesurfer recon-all viewed in fsleyes:

Many thanks
Yan

[neurolabusc]

@kiragy02 this is the expected behavior. Note the letters for the dimensions are correct.

You will want to set the FSLeyes display space to world space, rather than the atypical voxel order FreeSurfer uses. The are 48 orthogonal combinations. Tools like FreeSurfer Freeview, MRIcroGL and NiiVue will losslessly reorder dimensions to provide a roughly constant orientation, but FSLview’s voxel space is a truthful interpretation of the order of voxels saved to disk.

Since world space voxels and screen space pixels may not be aligned, you may observe aliasing artifacts. To reduce these, select no interpolation for your volume settings.

For more details, see this post.

[kiragy02]

Thank you so much for the information! I would greatly appreciate any troubleshooting suggestions for the misregistration by fmriprep.

Best wishes
Yan

[jsein]

@kiragy02, could you show the report for the anatomical part the see if the brain mak looks ok and if he contrast of the T1w is appropriate?