Hi FSL experts,
Sorry to repost, I am wondering if it is possible yet with the current FSL version to use a warp registration with FIRST and still be able to use the .bvars files to perform a surface shape analysis?
(see below)
thanks,
Kyle Kern
Hi,
I have been trying to use FIRST with FNIRT as described in another post to get better segmentations:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1003&L=FSL&P=R24263&1=FSL&9=A&J=on&d=No+Match%3BMatch%3BMatches&z=4
In the few subjects that did not segment well with FIRST, I have been able to get better meshes/volumes by finding the warpfield to MNI space using fnirt, running run_first on the image in MNI space (using the identity transorm for the -t option) and then putting the meshes back into native space (run_mesh_utils --doWarpMesh -i warp_field -m mesh.vtk -o out.vtk, [I was surprised to find that it needed the native-->MNI warpfield and not the inverse]), and also converting the mesh to a volume (first_utils --meshToVol -m mesh.vtk -i t1_image.nii.gz -l fill_value -o output_name).
However, is there a way to get the .bvars for the meshes back in native space since I will need these to perform shape analysis? I see the --MeshToBvars option in run_mesh_utils but I cannot figure out the proper way to use it. Or, alternatively, is there a way to run_first with a warpfield instead of an affine matrix so that the bvars created will list the warp field instead of the identity matrix? What is the best way to use fnirt with FIRST and still be able to perform the shape analysis?
thanks,
Kyle Kern
|