Dear Eini,
> Hi all!
>
> I was wondering what I'm doing wrong 'cause when I do coregistration
> (modality of target is T1 MRI and Object EPI) the SPM99 displays only grey
> matter and white matter segmentations but not CSF. SPM displays the mean,
> standard deviation and number of voxels also for CSF, but no actual images.
> Is there some default I should change or is there some kind of bug i've ran
> into? So far I haven't changed any defaults concerning coregister.
> Can this (missing CSF-images) affect the results of goregistration?
>
The object of the segmentation in the coregistration is to generate "segment
images" for images from both modalities. These are then smoothed and a
"simple" minimisation of sum of squared errors is then performed to register
the "segement images" to each other.
As you can well imagine one could probably get a quite good result just
registering for example gray-matter "segment images" from the two modalities
together. On the other hand it is concievable that slightly better results can
be obtained from a simultaneous registration of gray->gray and white->white
since additional information is included, hence this is what is being done in
SPM.
However, the additional inclusion of CSF will yield very little additional
information since the absence of white and gray matter simply indicates the
presence of CSF (i.e. the probabilities add up to one). For that reason SPM
does neither create or use a CSF map for the coregistration (although one is
calculated "behind the scenes").
Bottom line, you are doing nothing wrong (coregistration-wise) and need not to
worry.
Good luck Jesper
|