I am output data using MNI152T1NLin6Asym:res-2 and T1w. My T1w images are 256x256x168 and bold is 64x64x41. When I view the output space-T1w_desc-preproc_bold.nii.gz, the images are even lower than the input bold: 40x47x41. Why is the space-T1w image not in the proper dimensions?
Summary of what happened:
I have ran fmriprep v24.1.1 on a 2 session dataset. Each session has 1 T1w and 1 fMRI. The T1w images are already brain extracted so I set --skull-strip-t1w to skip and output spaces as MNI152NLin6Asym:res-2 and T1w. The anat report looks just fine (brain mask/tissue segmentation, MNI registration shows 2mm resolution, reconall). However, when I view the alignment of the functionals to the T1w, the T1w is a lower resolution than the native func space (native func is 64x64x51 while the resulting resolution is 40x40x41):
Command used (and if a helper script was used, a link to the helper script or the command generated):
fmriprep-docker \
rawdata \
derivatives \
--participant-label 1101 \
--skip_bids_validation \
--write-graph \
--no-tty \
--nprocs 4 \
--mem-mb 16384 \
--bids-filter-file rawdata/bids_filter.json \
--output-spaces MNI152NLin6Asym:res-2 T1w \
--skull-strip-t1w skip \
--fs-license-file .license \
-w fmriprep_work
Version:
v24.1.1
Environment (Docker, Singularity / Apptainer, custom installation):
Docker
Data formatted according to a validatable standard? Please provide the output of the validator:
PASTE VALIDATOR OUTPUT HERE
Relevant log outputs (up to 20 lines):
PASTE LOG OUTPUT HERE
Screenshots / relevant information:
Not an expert in this, but here is my take, from reading other threads. The T1w-space func image is aligned with your T1w image but it isn’t upsampled. The reason is to preserve quality. It is also smaller than your original, because for some reason, fmriprep removes the extra padding around the brain, shrinking the bounding box to the smallest dimensions possible. (This is quite annoying for me as the nominal resolutions won’t match between them and I have to transform whatever mask of atlas I want to use.)
So you haven’t lost any actual image quality, just empty voxels. If you want the images to match, I guess you need to either warp your image or ask fmriprep to output in T1w:res-2.
Perhaps this or this or this or probably this would help (ignoring the MNI reference)