Dear Aki,
> Hello, The formatting of my question was removed after posting. I've posted it below again, hopefully it will show cleanly this time. Thanks, Aki Hello FSL Experts, I am having some trouble using FNIRT for nonlinear registrations. It seems that fnirt is failing to use the affine transform from the betted structural image to the MNI_brain template- this seems to cause fnirt to fail when it attempts to register the non brain extracted image to the MNI template. I've attached the failed fnirt structural output, as well as a copy with the MNI 2mm template overlaid on the failed registration. I've pasted the exact code below for a better idea of what we're doing. We are using the betted hi res for the functional registration and first structural registration, and the non betted hi res for the fnirt. We are following the fnirt FSL practical pattern.
> #Func to HiRes /usr/local/fsl/bin/flirt -in ${Func} -ref ${betMPRAGEinput} -omat ${Func_Struct_Transform} -out FuncReg2HiRes.nii.gz -bins 256 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 6
> #Initial HiRes to MNI /usr/local/fsl/bin/flirt -in ${betMPRAGEinput} -ref ${RegistrationStandard} -omat ${Struct_MNI_Transform} -out HiRes2StructLinear.nii.gz -bins 256 -cost corratio -searchrx -90 90 -searchry -90 90 -searchrz -90 90 -dof 12
> #Whole brain Hi Res to whole brain MNI registration fnirt --iout=HiRes2Struct_Nonlinear.nii.gz --in=${origMPRAGEinput} --aff=${Struct_MNI_Transform} --cout=${NonLinear_Struct_MNI_Transform} --ref=${RegistrationStandardWholeHead} --config=T1_2_MNI152_2mm
as far as I can see these calls look good. Have you looked at your HiRes2StructLinear.nii.gz to see if flirt has done a good job?
Jesper
> Best Regards, Aki
|