In proper NIfTI images, the orientations are represented by a
voxel-to-world mapping, which maps indices of voxels to mm coordinates.
SPM can still read the old ANALYZE format images, but if it changes the
headers or writes any new images, then the format will be changed to
NIfTI.
So, if ANALYZE images are used initially, it is up to the users to
ensure that SPM is told what the correct orientation of these images is
(because any derived images will be in NIfTI format and the NIfTI
headers will contain the same information). The
spm_flip_analyze_images.m function is ignored once images are in NIfTI
format.
The Display button allows you to click on different parts of the image,
and it shows the mm coordinates (as well as voxel indices). SPM (and
other NIfTI compliant packages) assume that the right side of the images
has larger (or more positive) x coordinates (in mm).
To answer your question, the voxel-to-world mapping is encoded in two
ways: sform and qform. SPM primarily uses the sform, and generally sets
the qform to have the same values (the exception being those images that
have been imported for registration with Dartel). These fields are
documented at http://nifti.nimh.nih.gov/nifti-1
Occasionally, people notice what they believe to be a left-right mix up
when they use the slices option from the results. This is really a
feature rather than a problem, as the idea here is to show the original
slices in the image. For example, if the images are stored saggitally,
then saggital slices are displayed - even if the voxel-to-world mapping
indicates the correct transform to axial. Similarly, axial or coronal
slices are displayed according to how they are actually stored in the
image.
Best regards,
-John
On Thu, 2010-05-27 at 16:26 -0400, [log in to unmask] wrote:
> Dear John,
>
> In what variable of the NIFTI header is this info represented?
>
> Thanks,
> Eric
>
> Quoting John Ashburner <[log in to unmask]>:
>
> > SPM assumes that the image orientation is what the data format says it
> > is. If the data were to be correctly converted to NIfTI format, then
> > SPM should handle them perfectly fine. With the older ANALYZE format,
> > left-right orientations are not specified in the headers, so SPM needs
> > to be explicitly told what the orientations are.
> >
> > Best regards,
> > -John
> >
> >
> > On Tue, 2010-05-25 at 17:05 +0100, Stephen Paisey wrote:
> >> Dear Developers and users,
> >> We have imaged rats with a brain lesion on the right hand side on a
> >> Bruker scanner using Paravision 5.0. I have then used analyze 9.0
> >> to convert the raw bruker files into and img/hdr file pair. The
> >> images in analyze 9.0 are displayed the correct way around but when
> >> I load them in spm5 (i am using this version for its compatibility
> >> with the spmmouse extension) i get an image which has been left
> >> right inverted i.e. the lesion is now on the left side of the
> >> brain. I can get around this by flipping the image the wrong way
> >> around before i read it into spm but this is clearly a bug which
> >> could lead to totally wrong experimental results. It is lucky i had
> >> non symmetrical rats otherwise i probably wouldn't have spotted
> >> the error.
> >>
> >> Best Regards
> >>
> >> Stephen
> >>
> >
> > --
> > John Ashburner <[log in to unmask]>
> >
> >
>
>
>
--
John Ashburner <[log in to unmask]>
|