Hi
Elizabeth,

This can
be accomplished by two steps:

(1) Use “ivf”
to invert the displacement field (the *wrp.nii) outputted from
3dwarper.

(2) Use “applywarp3d”
to transform the ROI to the individual subject space.   

I will elaborate
on each of the above to steps below.

Let’s say that you use 3dwarper to transform a subject image “sub.nii”
to the standard space by matching it to a target image “trg.nii” using the
following command:

3dwarper –v –sub sub.nii –trg trg.nii –acpc –A

This command will create a “displacement field” called
sub_wrp.nii.   This is basically a 3D
vector field in nifti format.   It
encodes the nonlinear transformation needed to match to “sub.nii” to “trg.nii”.  

Now the problem you are faced with is to match an ROI which
is in the space of trg.nii to the space of the sub.nii image.   Therefore, you will need the inverse of
sub_wrp.nii.   So the first step would be
to invert  sub_wrp.nii using the “ivf”
program that has been written for this purpose.  

Let’s say that the subject image is of matrix size
(256x256x180) and voxel size (.94x.94x1.2) while the target image is of a
different matrix and voxel size.   The
command syntax for running ivf would be as follows:

ivf –nx 256 –ny 256 –nz 180 –dx .94 –dy .94 –dz 1.2 –i
sub_wrp.nii –o trg_to_sub_wrp.nii

The above command will create the new inverted displacement
field trg_to_sub_wrp.nii.   Note that you
needed to explicitly specify the size of the output displacement field “trg_to_sub_wrp.nii”
because the input displacement field “sub_wrp.nii” has the matrix and voxel
size of the target image which is different from the subject image.  

Finally, you can use applywarp3d to transform your ROI (say
it was called ROI.nii) to the space of the subject image using the following
command:

applywarp3d –nn –w trg_to_sub_wrp.nii ROI.nii

This will create the output file CROI.nii which will be in
the space of the subject image.   Note:
in the above command I used the –nn option for (nearest neighbor)
interpolation.   Otherwise, interpolation
effects will cause the transformed ROI to be fuzzy along the edges which may or may not be desirable depending on your application.

Hope this helps.

Babak