| Is there an easy way to transform the .mat files (SPMt.mat) from
| an analysis to stereotaxic space using the sn3d.mat file from a
| stereotaxically normalized target image?
| 
| Here's the situation:  We ran 12 separate spm analyses.  All of the
| fMRI data and the T1 image were co-registered, and the activations
| were plotted onto the T1 MRI.  The Statistical output, however, does
| not contain accurate stereotaxic coordinates, because the images
| were not stereotaxically normalized.  The result of each analysis 
| was saved as *spm.img and SPMt.mat files.  Yesterday, I
| stereotaxically normalized the T1 image, and selected the spm.img
| files as "other images," to put them in stereotaxic space.  However,
| the .mat files associated with these 12 analyses still reflect the
| old spatial orientation.   I want to "piggyback" all 12 of the .mat files
| into stereotaxic space like I transformed the spm.img files.  Is there
| an easy way to do this?  I tried the reslice only option, but this 
| routine wants me to enter .img files, rather than .mat files....

In theory it is possible to transform the results with an affine
transformation, and still have them as valid, but (in addition to
changing various 4x4 affine transformation matrices) there would be
lots of messing around with doing various transformations to the
smoothness estimates, and I think it would probably be a lot of work.

Transforming the results using nonlinear warps (based on the DCT basis
functions) would be much harder.

So, no easy way I'm afraid.

| Recently, someone asked how to plot activations onto a
| specific subject's own MRI data using the render function.
| The response was that 8 bit data was required.  Wishfully,
| I tried to Xtract a brain for rendering in spm99....
| without success.  The image was 16 bit.  Do I need to 
| convert this to 8 bit first, and repeat the Xtract, or is 
| there now a way to use 16 bit data for this step?  If I 
| need to convert it, shouldn't I be able to write out 8 bit
| data from within spm after loading the 16 bit image?

Many of the C routines have been re-written since SPM96, so they should
all support the same range of data types.

For the <Xtract Brain> function to work, you first need to have segmented
the images using the <Segment> button.  This gives the grey and white matter
images (as 8 bit) that are used by the brain extraction routine.  One thing
to note is that the brain extraction is very memory hungry, so is likely
to result in problems when run on a machine with limited memory.


All the best,
-John



%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%