fMRIPrep Issues with offset susceptibility distortion correction (SDC) on the EPI

Hi neurostars,

We have a subject that is running into an issue when running through preprocessing using fmriprep 1.3.2.
The resulting output of one of the task shows that the distortion correction did not run correctly. it looks like the resulting image is offset. We have reran this subject a few times through fmriprep (usually this fixes the issue when encountered on other subjects), but the results are the same.

/usr/local/miniconda/bin/fmriprep \
--fs-license-file /tmp/.license/freesurfer/license.txt \
--fs-license-file /tmp/.license/freesurfer/license.txt \
-w /tmp /tmp/BIDS /tmp/derivatives participant \
--participant_label 877168 \
--output-space template fsaverage5 \
--template MNI152NLin2009cAsym \
--template-resampling-grid native \
-n-cpus 16 \
--omp-nthreads 4 \
--mem-mb 64000 -v

Any advice on what to try next?


In case it helps: the affected runs (scans) vary across fmriprep runs (starting with the same images/settings/container/etc.).

For example, here are mean images for six runs from two different fmriprep instances.

The runs marked with red and blue switch from proper alignment to offset, while the other runs are correct in both instances. There doesn’t seem to be anything unusual about this participant or the affected runs (movement, signal quality, etc. all typical), or any way to predict these failures.