More distorted images are produced through SyN distortion correction

Summary of what happened:

Dear fMRIPrep experts,

I’m using SyN distortion correction to correct our resting-state functional images distorted in the AP direction because we do not have gradient-echo or spin-echo fieldmap images. However, I noticed that the “Corrected” images were oddly stretched. “Distorted” images appear rather undistorted. I attached the screenshot of “Corrected” and “Distored” images of a same subject from the fmriprep HTML report.
Other users have reported similar phenomena (SyN-based SDC, fieldmap-less estimation / More distorted images using --use-syn-sdc on fmriprep 23.0.1 than fmriprep 22.0.2), any updates on how to solve this problem?
Do I have to give up distortion correction? The original image may be less distorted, but of course it is not totally undistorted.

Best regards,

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

/opt/conda/bin/fmriprep /data /out participant --skip-bids-validation --participant-label XXXX --omp-nthreads 1 --bold2t1w-init register --bold2t1w-dof 6 --force-bbr --slice-time-ref 0.5 --dummy-scans 5 --random-seed 0 --project-goodvoxels --output-layout bids --medial-surface-nan --use-aroma --error-on-aroma-warnings --return-all-components --fd-spike-threshold 0.5 --dvars-spike-threshold 1.5 --skull-strip-fixed-seed --skull-strip-t1w force --use-syn-sdc --fs-no-reconall --stop-on-first-crash --fs-subjects-dir /opt/subjects -w /scratch --output-spaces MNI152NLin6Asym


fmriprep 23.0.0

Environment (Docker, Singularity, custom installation):


Screenshots / relevant information:

You do not need to give up on distortion correction, but I can’t promise that it’s fixed in the upcoming release. I’ll be spending more time on SyN in the new year.

Dear @effigies ,

Thank you for your kind reply.
I understand the situation. I look forward to future updates.
Please take good care of yourself!

Sincerely yours,

Hi @Printemps ,

For the time being, you may want to test Synbold-DISCO and the possibility to use it with fmriprep as it is discussed here: Using synthetic field map from synbold-disco with fmriprep