Because it is based on minimising the mean squared difference, the old
spatial normalisation routine only works for images with the same
sorts of intensities as those of the template. If your EPI images
with contrast agent have a different pattern of intensity from the EPI
template, then things are not going to work very well.
If you have EPI images collected at the same time, that have similar
intensities to the EPI template, then you could maybe try estimating
the spatial transformation from these instead. Alternatively, you
could try spatially normalising via the segmentation (as in the
manual).
Best regarsd,
-John
On 25 October 2012 13:19, Felix Schlegel <[log in to unmask]> wrote:
> I encountered the same problem and I'm wondering whether anyone has come up with an SPM solution.
> When I try to normalize the EPI images after injection of the contrast agent (Endorem) to the baseline scans before injection, the result looks completely distorted.
>
> These are the parameters I used, maybe something could be changed here? The contrast agent EPIs are much darker, but not more distorted. Also the brain is almost exactly in the same position anyway, so is there a way to further penalize large stretching/warping?
>
>
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.smosrc = 3.75; (corresponds to 1.5x the voxel size)
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.smoref = 0;
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.regtype = 'subj';
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.cutoff = 25;
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.nits = 16;
> jobs2{1}.spm.spatial.normalise.estwrite.eoptions.reg = 10;
> jobs2{1}.spm.spatial.normalise.estwrite.roptions.preserve = 0;
> jobs2{1}.spm.spatial.normalise.estwrite.roptions.bb = [NaN NaN NaN;NaN NaN NaN];
> jobs2{1}.spm.spatial.normalise.estwrite.roptions.vox = [2.63, 2.33, 5];
> jobs2{1}.spm.spatial.normalise.estwrite.roptions.interp = 4;
> jobs2{1}.spm.spatial.normalise.estwrite.roptions.wrap = [0, 0, 0];
|