You may want to add an additional component to the job, (by selecting
"Identity (reference image)".  This should make sure that the inverse
warped images are written out with the same dimensions etc as the
reference image that you select.

You would use the  seg_inv_sn file, and if the idea is to warp t
statistic images back to native space, you'd specify the spmT images
that you hope to inverse warp.  I would suggest leaving the bounding
box and voxel sizes as they are.

Overall, the batch (if saved as a .m file) should have the following structure:

matlabbatch{1}.spm.util.defs.comp{1}.sn2def.matname = {'image_seg_inv_sn.mat'};
matlabbatch{1}.spm.util.defs.comp{1}.sn2def.vox = [NaN NaN NaN];
matlabbatch{1}.spm.util.defs.comp{1} = [NaN NaN NaN
                                                  NaN NaN NaN];
matlabbatch{1}.spm.util.defs.comp{2} = {'image.nii,1'};
matlabbatch{1}.spm.util.defs.ofname = '';
matlabbatch{1}.spm.util.defs.fnames = {'wimage.nii,1'};
matlabbatch{1}.spm.util.defs.savedir.savepwd = 1;
matlabbatch{1}.spm.util.defs.interp = 1;

Best regards,

On 10 May 2011 17:22, Marine Lazouret <[log in to unmask]> wrote:
> Dear SPMers,
> We would like to extract the activation clusters for each subject in its own
> native space.
> I need to transform back the activation clusters from MNI to native space.
> I tried using the deformation batch but I'm not sure which images to use in
> "Apply to", the originally normalized images?
> For the parameter file I should use the seg_inv_sn file right?
> Should I keep the default parameters for bounding box and voxel size?
> I added the screen shot of the batch to make myself more clear,
> Best Regards,
> --
> Marine