Issue with subcortical alignment from BOLD to T1

Summary of what happened:

Hi all, I’m hoping to get some suggestions for how to handle alignment in my dataset. I have 3T BOLD images for a set of participants whose structural scans were collected at 7T. When I run fmriprep, the cortex seems to align well between BOLD and T1w images, but everything subcortical looks shifted. The brainstem looks misaligned, and the cerebellum as well. I’ve tried several things and can’t seem to resolve the problem. I’ll post the general code that I’ve used below with a list of different commands I’ve tried.

Things I’ve changed about this:

  1. I’ve tried running without use-syn-sdc
  2. I’ve tried adding --bold2anat-init header
  3. I’ve tried adding --bold2anat-dof 9

None of these seem to make significant changes in the alignment between my BOLD and T1w data.

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

while read SUBJ; do
    echo "============================="
    echo " Running fMRIPrep for $SUBJ "
    echo "============================="

	# make a subject-specific temp dir
        mkdir -p $fmriprep_tmp

	singularity run --cleanenv \
	  -B $bidsdir \
	  -B $fmriprep_tmp \
	  -B $cache \
	  -B $fmriprep_output \
	  -B ${fs_lic}
	  $fmriprep \
	  $bidsdir \
	  $fmriprep_output \
	  participant \
	  --participant-label $SUBJ \
          --task-id rest \
	  --work-dir $fmriprep_tmp \
	  --output-spaces T1w MNI152NLin2009cAsym \
	  --ignore slicetiming \
	  --use-syn-sdc \
	  --fs-license-file /opt/freesurfer/license.txt \
          --derivatives $anat \
          --skull-strip-t1w skip \
          --fs-subjects-dir $freesurfer \
          --fs-no-reconall \
	  --skip-bids-validation

	echo "Finished $SUBJ"
done < "$subj_list"

echo "All subjects processed."

Version:

fmriprep-25.2.3

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

Singularity

Data formatted according to a validatable standard? Please provide the output of the validator:

BIDS validated dataset -- no issues that would affect preprocessing

Relevant log outputs (up to 20 lines):

No errors to report

Screenshots / relevant information:


Hi @kylieisenburg,

A picture might help us understand your problem better. You can try SynBOLD-DisCo if you don’t have fieldmaps, and try re-enabling FreeSurfer / surface based registration workflows.

Best,
Steven