Hello,
If you are seeing edge-effects in your warped mask, then binarising ( or thresholding then binarising ) the output mask in fslmaths may do what you want.
Kind Regards
Matthew
--------------------------------
Dr Matthew Webster
FMRIB Centre
John Radcliffe Hospital
University of Oxford
> On 9 Nov 2017, at 22:46, SUBSCRIBE FSL Anonymous <[log in to unmask]> wrote:
>
> Hello,
> I registered T1 with lesion in standard space MNI, I using flirt and fnirt, the output of T1-registration and the mask-registration has wavering effect from one slice to another, is there any reason why that happens and a solution to the problem?
>
> I tryed three tests:
> 1. the data inserted in Flirt and Fnirt are isotropic 1x1x1 and once I registered T1 it didn't seen to have any problem but the registered mask had an intensity of less than 1 on the edges.
> Commands
> flirt -in T1file -ref T1file -out xxx -applyisoxfm 1
> fnirt --ref=MNI152_T1_1mm --in=T1file --inmask=invmask --aff=matrix --iout=xxx --fout=xxx
> applywarp -i file -o xxx -r MNI152_T1_1mm -w filebyfout --interp=nn
>
> 2.I addition to the first attempt I added the interpolation nn command to the command that makes isotropic data 1x1x1 but a problem comes up, when I register the T1 and the mask i get a wave effect (moving from one slice to the other).
> Commands
> flirt -in T1file -ref T1file -out xxx -applyisoxfm 1 -interp nearestneighbour
> fnirt --ref=MNI152_T1_1mm --in=T1file --inmask=invmask --aff=matrix --iout=xxx --fout=xxx
> applywarp -i file -o xxx -r MNI152_T1_1mm -w filebyfout --interp=nn
> I think this is wrong because I just want the binary mask.
>
> 3. I use interpolation only in commands for mask, but there is always the wave effect.
> flirt -in T1file -ref T1file -out xxx -applyisoxfm 1
> flirt -in maskfile -ref maskfile -out xxx -applyisoxfm 1 -interp nearestneighbour
> fnirt --ref=MNI152_T1_1mm --in=T1file --inmask=invmask --aff=matrix --iout=xxx --fout=xxx
> applywarp -i file -o xxx -r MNI152_T1_1mm -w filebyfout --interp=nn
>
> Thank you for your help.
> S
|