Summary of what happened:
Does fMRIPrep support lesion masking in T1w space without freesurfer reconstructions? I added a lesion into my raw BIDS dataset under the anat folder, named it like:
sub-64_label-lesion_roi.nii.gz
sub-64_label-lesion_roi.json
However, the lesion mask was not taken into account? Is it because freesurfer is not used?
Coregisteration issues
Also, the patient being analyzed here has a massive lesion and some hemispheric asymmetry (the damaged hemisphere is much smaller), which is why coregisteration is not consistent across runs. For example, there are variabilities between run to run coregisteration, the orientations and angles of the boundaries are different.
One run (faulty) – note the tilt.
Another run (looks correct)
I did not run into any issues related to coregisteration in SPM, however I ran into issues in segmentation. On the other hand, segmentation here is good but coregisteration is not.
All analyses will be conducted in native space however, MNI normalization transforms will be used because I will inverse normalize ROIs.
Also, does using a lesion mask have any implications for coregisteration? For example, instead of coregistering using cerebral features, can one not use the overall skull and outlines and rougher margins, instead of the small details?
Command used (and if a helper script was used, a link to the helper script or the command generated):
fmriprep ./bids ./bids/derivatives/fmriprep_lesion participant --participant_label 64 --work-dir ./fmriprep_work_lesion --output-spaces T1w --write-graph --skip-bids-validation --skull-strip-t1w force --fs-license-file license.txt --fs-no-reconall --bids-filter-file task_filter.json
Version:
fMRIPrep v25.2.5