> > I'm referring to myself of course...
>
> Apologies if this has been raised before but I spent a deal of
> time surfing the mailbase for a clear answer and was not too
> successful.
>
> Just to summarise, there are two issues regarding image orientation
> when moving from spm99 to spm2b: normalisation and display defaults.
>
> The code for switching on/off the neurological/radiological flip during
> spatial normalisation is in spm_flip_analyze_images.m and the value
> of variable flip specifies whether or not flipping occurs: 1=yes,
> 0=no;
>
> Display. I had a problem when comparing an image using Display in
> spm99 then spm2b - it looked like it was being flipped in spm2b.
>
> For spm2b, the file spm_defaults.m has this bit:
>
> % File format specific
> %==================================
> defaults.analyze.multivol = 0;
> defaults.analyze.flip = 1; << very important relates to L/R
>
> With the default value of 1, the image was flipped, with 0
> it's not.
>
> Again, sorry if this is becoming tiresome but it nearly did my melon in
> today! John, you must feel like Aristotle trying to persuade the
> learned that world is round...
One day this will seem really easy.
Analyze images are stored in a left-handed co-ordinate system.
Talairach co-ordinates use a right-handed system. In SPM99, the
images were flipped during spatial normalisation in order to get
from a left-handed to a right-handed system. In SPM2, no flipping is
necessary, because it is told about the handedness to expect from
Analyze images (if defaults.analyze.flip==1, then it is left-handed,
if defaults.analyze.flip==0, then it is right-handed). See:
http://www.fil.ion.ucl.ac.uk/spm/spm2.html#Compat
defaults.analyze.flip is used if there are no SPM2 format .mat files
associated with the images (i.e. if there is no .mat file, or if
the .mat file only contains an "M" variable, but does not contain a
"mat" variable). It is used to tell SPM whether there should be any
reversal of handedness in the mapping from voxel-to-world co-ordinates.
If there is a reversal of co-ordinate system, then SPM2 will interpret
the co-ordinate system as flipped relative to SPM99, and display the
images accordingly.
I wanted to avoid the flipping because:
1) Spatially normalised images from SPM99 are not in a proper Analyze
format (because they are stored with a right-handed system).
2) I wanted the same rendering utilities etc to be applicable to the
un-normalised as well as the normalised images. For example, if
someone did a brain extraction and rendering with non-normalised
data in SPM99, then the images would be mirror images. In SPM2, the
rendered images should not be mirror images.
All the best,
-John
--
Dr John Ashburner.
Functional Imaging Lab., 12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491 or +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420 http://www.fil.ion.ucl.ac.uk/~john
|