Do you have all the updates and fixes for SPM5b installed? If so, then it
could it be possible that SPM5b has converted the header to NIFTI format - in
which case SPM5b will read the orientation from the header - ignoring the
.mat file. If there is a .mat file, then SPM2 will read the orientation from
this as it is unable to obtain this information from the header. If there is
no .mat file, then the behaviour of SPM2 with NIFTI format images is rather
unpredictable.
If these suggestions don't help, then you could email me the .hdr and .mat
file so I can try to figure out what is happening.
Best regards,
-John
On Saturday 17 September 2005 21:18, Ray Fix wrote:
> Thank you for your response. Sorry for not replying sooner as I was
> away from my email.
>
> If I use the same exact input files (img/hdr/mat) with SPM2 and SPM5b
> the direction cosines are different. So, while it would appear that
> SPM5b can still read analyze files, it ignores the .mat file. Is
> this the intent? My guess is that everyone running SPM5b is happily
> using Nifti and no one is testing Analyze data. I could be wrong
> though.
>
> Best regards,
> Ray
>
> PS: I am fully on board with using NIFTI in the future. It avoids a
> lot of potential mixups and conversion mistakes.
>
> PPS: For the record, FreeSurfer's mri_convert makes fully compliant
> SPM Analyze format (with mat files).
>
> On Sep 14, 2005, at 7:16 AM, John Ashburner wrote:
> > In the standard Analyze format, there is no space allocated in the
> > headers for
> > the orientation of the images. This is why additional .mat files
> > are used by
> > SPM99 and SPM2. If the converter doesn't create such .mat files,
> > then the
> > direction cosines will simply be the default values. One of the
> > features of
> > NIFTI is that space in the headers has been used in order to store the
> > orientations of the images.
> >
> > Best regards,
> > -John
> >
> >> I am converting to both Analyze and Nifti format with the freesurfer
> >> tool mri_convert (from COR format).
> >>
> >> When I convert to Analyze and display in SPM5b, the head is flipped
> >> around in a funny way and the dir cos matrix is identity.
> >>
> >> When I convert to Nifti and display in SPM5b, the head is in what
> >> seems to be correct orientation with dir cos [1 0 0 ; 0 0 1 ; 0 -1
> >> 0].
> >>
> >> When I display Analyze format in SPM2, it looks like the Nifti
> >> display of SPM5b and the dir cos is also [1 0 0 ; 0 0 1; 0 -1 0].
> >>
> >> Are the dir cos just a function of display? Should I be worried
> >> about this?
|