Dear FSL expert,
I asked a question regarding registration residual around ventricles using flirt registration from T12b0. Briefly, I attempted to register T1 ROI masks to b0 space with flirt, everything works fine except the frontal horn of ventricles.
Here is the link https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1703&L=FSL&P=R36410&1=FSL&9=A&I=-3&J=on&K=1&X=B8D013C1C058CDCFC3&Y=joyce.yu.guo%40gmail.com&d=No+Match%3BMatch%3BMatches&z=4
Emmanuel suggested to use fnirt, So I followed guides from fslwiki https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FNIRT/UserGuide to register both b0 and T1 to standard space.
Command lines are:
T12std
flirt -ref ${FSLDIR}/data/standard/MNI152_T1_2mm_brain.nii.gz -in T1_brain.nii.gz -omat affine_T12std.mat
fnirt --in=T1.nii.gz --aff= affine_T12std.mat --cout=nonlinear_T12std.nii.gz --config=T1_2_MNI152_2mm.nii.gz
applywarp --ref=${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz --in=T1.nii.gz --warp=nonlinear_T12std.nii.gz --out=warpedT12std.nii.gz
applywarp --ref=${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz --in=GM.nii.gz --warp=nonlinear_T12std.nii.gz --out=warpedGM2std.nii.gz
applywarp --ref=${FSLDIR}/data/standard/MNI152_T1_1mm.nii.gz --in=WM.nii.gz --warp=nonlinear_T12std.nii.gz --out=warpedWM2std.nii.gz
FA2std
flirt -ref ${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain -in FA.nii.gz -omat affine_FA2std.mat
fnirt --in=FA.nii.gz --aff=affine_FA2std.mat --cout=nonlinear_FA2std.nii.gz --config=FA_2_FMRIB58_1mm
applywarp --ref=${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain.nii.gz --in=FA --warp=nonlinear_FA2std.nii.gz --out=warped_FA2std
applywarp --ref=${FSLDIR}/data/standard/FMRIB58_FA_1mm_brain.nii.gz --in=b0.nii.gz --warp=nonlinear_FA2std.nii.gz --out=warped_b02std
I wonder if this pipeline is OK? I do have ${FSLDIR}/data/standard/FMRIB58_FA_1mm image without suffix '_brain', but there is no skull in this image. So, they should be the same?
The warped_FA2std.nii.gz aligned perfectly with MNI152_T1.nii.gz, but when I applywarp 'nonlinear_FA2std.nii.gz' on b0, the problem of 'mis-registration residuals' around ventricles remains. I wonder if there is other ways to correct it? I wonder if there is a way to reverse transformation filed from FA2std 'nonlinear_FA2std.nii.gz' to std2FA. So that, I can apply warp filed std2FA on standardized masks back to b0 space? Any suggestions are appreciated !!!!
Best Wishes,
Joyce
|