> We have aquired functional data that is oblique with respect to the
> axial plane. SPM seems to be able to analyze for function, but now
> I want to plot on top of structural image. First, the volumes are not
> full brain and the planes of acquisition are NOT axial thus I am not
> normalizing to talairach coords. However if I do "slices" I presume that
> the coordinates that SPM is using to pick the right slice still are
> valid for my structural image, which at the moment is just one of my
> functional images.
>
> Eventually I would like to plot to higher res (t2) structural images
> that we took with the same acquisition plane orientation. Will SPM do
> this for me automatically?
SPM96 will do this for you - providing that your structural images are
acquired parallel to your functional images, and that you have the correct
ORIGIN and VOXel sizes set in all your headers.
> Or should I worry that SPM has done
> transformations behind the scene that I am not aware of?
The transformations it will do are dependant on values for VOX and ORIGIN
in the .hdr files.
> (I have done at
> least the realing on the data set, though not the higher res t2
> structural images. The realignment is only a few mm in each direction so
> that should be ok if I am just looking for gross spatial resolution of
> function.)
>
> Finally, I will want to plot onto our own t1 structural data that is
> indeed in the normal axial plane. Presuming I can get the angle from the
> scanner data, I should be able to rotate the functional volumes into the
> correct orientation (something like 15-20 degrees).
You could use the co-register button to do this in SPM96, and superimpose
your results on to a structural image that has been resliced to match the
functional images. Unfortunately, the reslicing will bring the resolution
of the structural images to the same as that of the functional images
(which defeats the object of having high resolution structurals).
In theory, the high res images could be resliced to preserve the high
resolution, but this would require additional Matlab coding. This will
be more straightforward in SPM99 because information from the ".mat" files
can be used to reslice the images on-the-fly in the results section in
such a way that the resolution is preserved. This can not be done in
SPM96, since the ".mat" files are only ever used by the spatial
normalization, realignment and coregistration buttons, and are not used
in the results section. In SPM96, only the VOX and ORIGIN fields of the
.hdr files are used by the results section, so relative rotations can not
be included in on-the-fly resampling.
> Can you tell me what
> SPM functions I can use to do this rotation and reslicing? Does this
> sound like the correct approach.
Be careful when you register structural images with distorted functional
images. The registration is never that good (because of the distortions).
Personally, I prefer to see results superimposed on the functional images
themselves. This way you can see exactly where the activations have arisen
from (unless the functional images have been distortion corrected).
Regards,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|