What you have done seems to be correct. If you specify the bounding box as
nan(2,3) and the voxel sizes as nan(1,3), then the inverse normalised image
should be closer in dimensions and voxel sizes to the image that you
estimated the warps from.
Best regards,
-John
On Tuesday 05 August 2008 17:01, Wang, Ling wrote:
> Dear All,
>
> I was wondering if I did the inverse normalisation correctly. I use unified
> segmentation to do the normalisation. Now if I want to deproject spmT
> images to individual T1 images, which I use as the source image in the
> segmentation, what I need to is to use "Normalise: Write" with
> *seg_inv_sn.mat (produced in segmentation step) as "Parameter File" and
> spmT images as "Images to Write". Then the same coordinates (in world
> space) in T1 images and inverse normalised spmT images should indicate the
> same brain areas (almost), although they have different dimensions and
> origins.
>
> Thank you!
>
> Best wishes,
> Ling
|