Hey all,
I'm trying to implement this new BBR registration approach, but the results we are getting look really bad in pretty much all of our subjects, e.g. some EPIs hardly need moved at all to align them, but the script is moving them way off the brain.
The code I'm using is below. It's not clear to me whether the EPI needs to be skull-stripped or not, so I have tried it both ways, and they both seem off. I am using fsl_anat to get all of the structural images: biascorr, biascorr_brain, and biascorr_brain_wm_seg (which is just the output from FAST, T1_fast_pve_2). The EPI is slice-time, motion and field-map corrected before this step, but otherwise is unchanged. Also note that I have examined the anat_biascorr_brain files, as the skull-stripping is done using FNIRT, and they all look good.
Any thoughts? Your help is greatly appreciated. Thanks!
echo
echo "Coregistering EPI to anatomical..."
cd ${subjectsDirectory}/${subj}
rm __tmp_coreg_slice*.nii.gz
3dcalc -a ${subj}_resting_moco_fm_ss.nii.gz'[90]' -prefix __tmp_coreg_slice.nii.gz -expr "a"
epi_reg --epi=__tmp_coreg_slice --t1=${subj}_anat_biascorr --t1brain=${subj}_anat_biascorr_brain --out=${subj}_e2a
|