Dear FSL users,
I was trying to coregister post- and pre-operation T1 scans using fnirt (FSL version 4.1.4). I did it in the following way:
1. manually draw the resected brain area on post-op scan, and invert the resection mask (resected area is 0, anywhere else is 1)
2. using FLIRT to register two brains, and get the linear transform matrix
3. executing the command:
fnirt --in=post-op_T1 --ref=pre-op_T1 --aff=linear_transform_matrix --cout=warpfield --warpres=8,8,8 --refmask=pre-op_T1_mask --inmask=resection_mask_inverse --applyrefmask=1 --applyinmask=1
4. using applywarp to warp the resection mask to pre-op space
It ran smoothly without any error or warning message. However when I checked the warped resection mask, it was much smaller than it on the post-op scan (warped_postop_mask:17061 mm3 vs postop_mask:26746mm3), while the preop scan had a bigger intra-cranial volume (preop:1269678 mm3 vs postop:1192340 mm3)
It seems that the fnirt still trying to 'fill in' the resection area by pulling normal tissues nearby. Anyone knows how to overcome this issue? Any suggestion is very much appreciated!
Thanks,
Hugh
|