Hi experts,
Sorry to be of bother but I'm having a bit of a problem combining a flirt matrix with a fnirt warp.
I am trying to do a longitudinal VBM in stroke patients, following Gwaenelle Douaud's very helpful guidelines here:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1012&L=FSL&D=0&1=FSL&9=A&J=on&X=3E0AD22A0EF20AC0DE&Y=froudiss%40tcd.ie&d=No+Match%3BMatch%3BMatches&z=4&P=227997
I am currently at stage 6) "convolve the subsequent warpfield with the halfway .mat to obtain the final warpfield to apply to each native GM image" and am having a bit of trouble getting the combined warp to output acceptable results.
The two part registration works well to my eye, but when I try to combine the linear and non-linear transforms into one step, either using the following commands:
convertwarp -r ${FSLDIR}/data/standard/tissuepriors/avg152T1_gray.hdr -m TP1_to_halfway.mat -w GM2MNI_Non_Lin_transf -o TP1_GM_to_MNI_warp.nii.gz
applywarp --ref ${FSLDIR}/data/standard/tissuepriors/avg152T1_gray.hdr --in=TP1_GM --warp=TP1_GM_to_MNI_warp --out=TP1_GM_to_MNI2mm --interp=spline
OR just this command:
applywarp --ref ${FSLDIR}/data/standard/tissuepriors/avg152T1_gray.hdr --in=TP1_GM --premat=TP1_to_halfway.mat --warp=TP1_GM_to_MNI_warp --out=TP1_GM_to_MNI2mm --interp=spline
it results in some kind of binary mask, where only about 1/3 of the grey matter voxels are present (with an intensity of 1) and all other voxels equal to zero.
Any help in identifying my mistakes and explanation as to why the 2-part registration but not the 1-part registration works would be greatly appreciated.
Many thanks in advance,
Sean
|