Summary of what happened:
Dear QSIprep team,
thanks for the great software! We’re currently preprocessing our dataset using qsiprep-1.0.2
, and I’ve got a question about the anatomical processing workflow. Here’s an example of our derivatives anat folder:
ses-3
|-- anat
| |-- sub-1554_ses-3_from-ACPC_to-MNI152NLin2009cAsym_mode-image_xfm.h5
| |-- sub-1554_ses-3_from-ACPC_to-anat_mode-image_xfm.mat
| |-- sub-1554_ses-3_from-MNI152NLin2009cAsym_to-ACPC_mode-image_xfm.h5
| |-- sub-1554_ses-3_from-anat_to-ACPC_mode-image_xfm.mat
| |-- sub-1554_ses-3_from-orig_to-anat_mode-image_xfm.txt
| |-- sub-1554_ses-3_space-ACPC_desc-aseg_dseg.nii.gz
| |-- sub-1554_ses-3_space-ACPC_desc-brain_mask.nii.gz
| |-- sub-1554_ses-3_space-ACPC_desc-preproc_T1w.json
| |-- sub-1554_ses-3_space-ACPC_desc-preproc_T1w.nii.gz
| `-- sub-1554_ses-3_space-ACPC_dseg.nii.gz
However, we also have T2w, FLAIR and lesion masks in the subject space (registered to the original T1w), and we’d like them to be brought into AC-PC space as well. I tried running antsApplyTransforms with the transformation matrices provided by QSIprep, but this resulted in somewhat misaligned images:
antsApplyTransforms -d 3 -e 0 -i sub-1554_ses-3_T2w.nii.gz -r sub-1554_ses-3_space-ACPC_desc-preproc_T1w.nii.gz -o sub-1554_ses-3_space-ACPC_desc-preproc_T2w.nii.gz --interpolation Linear -t sub-1554_ses-3_from-anat_to-ACPC_mode-image_xfm.mat
According to this thread, QSIprep registration workflow is more complicated than suggested by the derivative matrices, and the easiest option is to run antsRegistration from scratch.
I was wondering, as of 2025, is there a more appropriate way to align all of the source anatomical images to the derivative space? Maybe giving QSIprep a list of files to tag along with T1w?
Thanks in advance!
Warm regards,
Victoria