Hi Guido,
> Here is the output for two files, both have been reconstructed
> from the same *.IMA file.
I'm not quite sure what you mean by two images reconstructed from the
same file. How do they differ? Have you used two different
reconstruction tools? Or processed one differently from the other?
> ------ SPM5, file 1
> -6.0000 0.0000 0.0000 66.0000
> 0.0000 5.8333 0.0000 -75.8333
> -0.0000 0.0000 5.8333 -75.8333
> ------ SPM5, file 2
> -5.9999 0.0058 -0.0309 66.2633
> 0.0060 5.8333 -0.0108 -75.6325
> -0.0318 0.0109 5.8332 -75.9835
So these matrices differ, in both the translation components and the
linear part. The differences are far greater than the limit of machine
precision, which I think implies either:
- you have manually reoriented the second file
- you have coregistered the second file
- file2 was created with a tool which was able to take rotations from
the ima file, while file1 just looked at origin and voxel-sizes
- or something else that I'm not clever enough to guess...
> ------ SPM2, file 1
> 1.0e+05 *
> -0.0001 0 0 0.0307
> 0 0.0001 0 -1.2843
> 0 0 0.0001 -0.0271
> ------ SPM2, file 2
> 1.0e+05 *
> -0.0001 0 0 0.0307
> 0 0.0001 0 1.4635
> 0 0 0.0001 -0.0348
SPM2 appears also to have got different origins for these two files,
but sadly, due to the 1.0e+05* factor, we can't see if it also has
some rotation in the linear part. (you could look at v.mat(1:3,1:3) to
see this more clearly.)
From the other info in your email it seemed that both files and both
version of SPM gave the same voxel-dimensions (to 5s.f.), which sounds
good enough to me (I think the difference is perhaps a very slight
rounding error resulting from the difference between
including/excluding rotation). So, either tell spm_check_orientations
to relax a bit, as you suggested before, or work out why the second
file appears to include a rotation. I don't think I can help further
without knowing the way these different files were created/treated.
Best,
Ged.
P.S. Please reply via the list, rather than to me directly -- someone
else out there might have some better ideas!
|