Dear Hanna,
> we transform our T1-images to standard-space with the MNI152_T1_1mm-
> image as
> reference. Then we use these transformed images to create our masks
> (parts
> of Gyrus precentralis and...)
> To transform these masks into diffusion-space, we use TBSS, our
> dti_FA-images and the FMRIB58_FA_1mm-image to create the warp and
> then we
> transform the masks with the inverted warp into diffusion-space.
>
> Our problem is, that sometimes parts of the transformed masks are
> not inside
> the brain.
there are many steps here. Let me see if I get this right.
What you want as an end result is masks defined in the individual
subjects T1 scan in the space of the same individual subjects dti
images. Right?
If so, what is the reason for your rather roundabout path? In your
method you have two non-linear transforms T1->MNI152 and FA->FMRIB58,
that will both have imperfections, and the implicit assumption that
MNI152 is identical to FMRIB58, which will only be approximately true.
The combination of these three sources of error I would guess will add
up to several mm in parts of the brain.
A better strategy, if I have understood you correctly, would be to
acquire fieldmaps with your dti data to allow you to correct the
distortions in them. That will also have errors, but I would imagine
those would be smaller than with your current strategy.
> I used Flirt and Fnirt to transform the MNI152_T1_1mm with
> FMRIB58_FA_1mm as
> reference but the output is not very satisfying.
fnirt would not be much help here I am afraid. The contrast is too
different in the FA and the T1 templates for fnirt to cope with.
Good Luck Jesper
>
> Can you help me?
>
> Thanks,
> Hanna.
>
>
>
>
>
|