hello all,
ive encountered an issue with regards to image orientation in spm 5 (on solaris w/ matlab 7). in doing an analysis across subjects using pet data in analyze format, i encountered this error:
Running "Factorial design specification"
Mapping files : The images do not all have same orientation and/or voxel sizes.
[2 0 0 -92; 0 2 0 -128; 0 0 2 -74] /data/3117_activation.img
<-- cut here for brevity -->
[-2 0 0 92; 0 2 0 -128; 0 0 2 -74] /data/3730_activation.img
??? Error using ==> spm_check_orientations
The orientations etc must be identical for this procedure.
though i cropped the above error message, there are many subjects with each of those two orientations. this looks to me like some of them were 'mirrored' in the x dimension (left/right). using the display function in spm 5, the subjects will show a voxel size of 2x2x2 OR -2x2x2 (in mm^3, of course), corresponding to the orientations shown in the error message. also using the display function, i can do a transformation by doing a "resize" of x by -1. this will allow me to make sure the orientations agree and then the analysis will complete. however, this causes the images to show up reflected, and i cannot be sure that all of the subjects line up left-to-right. this would be a much easier problem to solve if the image transformation was in any dimension other than x. so much for symmetry... :(
the possible bug comes into play because using spm2, all of the above subjects show a voxel size of -2x2x2!!! at first i thought our data was somehow inconsistent (though it may be, or at least our headers may not exactly conform to the specification) but now i am lost. any ideas? any help would be greatly appreciated.
cheers,
-george
|