Hi Corina,
Most likely Steve is right that this is a bad idea, but if you did
want to investigate masking out the lesion with nreg, then the
relevant option is:
<-Tp value> Padding value in target image
and voxels with that value will then be ignored in the similarity
computation.
E.g. if you have a binary mask image with 1=include and 0=exclude (as
suitable for e.g. flirt), then you'd want to do something like:
fslmaths msk -sub 1 msk_pad
fslmaths img -msk msk -add msk_pad img_pad
and then use:
-Tp -1
in nreg with img_pad.
If your original data can have negative values, check if it includes
-1, in which case you'll need to multiple your msk_pad by a value
which doesn't occur elsewhere in the image, and then use this with -Tp
(this is a bit of a pain, but there is no direct masking option).
Best,
Ged.
> We have been trying to perform some TBSS analyses on brain lesioned
> patients. Unfortunately, nreg does not perform well. We are trying to
> mask out the lesion (in the worst case we can use only the contralateral
> hemisphere to the lesion). However, it is not clear how the syntax for
> nreg should be. Also, should we use the x, y, z values in Tailarach
> space (0, 0, 0 @ AC) or in the original space (0, 0 , 0 @ left inferior
> posterior corner)?
|