> I've uncovered another piece of the puzzle. When I run a
> normalization using the SPM T1 template, after normalization the subject's
> image is displayed as right=right, and SPM states that the image is
> flipped. This is correct, right?
This is the usual behaviour.
> However, when I use a different template,
> namely a skull-stripped one, SPM doesn't do the flip, despite the fact that
> I set the defaults the same both times. In medx, the image header
> information for orientation is the same for both templates - left,
> posterior, superior. However, the skull-stripped template I'm using also
> contains a lot of mensuration information - could this be read by SPM?
I don't understand either. The only information SPM uses is that from the
.hdr files and .mat files. Note that if there is a .mat file for your
template, where the variable M has a negative determinant, or if an odd
number of the voxel sizes in the .hdr are negative, then this will indicate
a flip. When combined with the flip from the affine transformation, then
you have two flips, which cancel out.
best regards,
-John
> Subject: Re:
> > Oops! A typo on my part. I set the SPM99 defaults so that my initial
> > estimates are reflict my "radiologically" oriented images, meaning that
> > they should be [0 0 0 0 0 0 -1 1 1 0 0 0]. However, after SPM99 performs
> > the normalization, it states that no flipping was used, and my images
> > appear with right=left (instead of correctly, right = right; presumably
> > because no flip was done...)
>
> Is it possible that you reset the defaults, or restarted SPM? If this is
> done, then the defaults from spm_defaults.m are re-loaded.
>
> Best regards,
> -John
>
> > Subject: Re:
> > > I have a set of Analyze images in radiologic (R=L) format.
> > >
> > > I want to spatially normalize these to a template in neurological (R=R)
> > > format.
> > >
> > > Spm_sn3d.m states, "The normalized images should use neurological
> > > orientation for their orientation"
> > >
> > > So, I carefully choose my normalization defaults so that the initial
> > > estimates for the affine transformation is radiologic [0 0 0 0 0 0 -1
> > > -1
> >
> > -1
> >
> > > 0 0 0]
> > >
> > > However, when I run the normalization, the resulting display states,
> > > "Linear affine component - no flipping" - and the images appear
> > > left=right (which I presume to be incorrect, since SPM told me at this
> > > state everything should be in neurologic format)
> > >
> > > Can someone please shed some light on this? I've sifted through the
> > > archives, and while there is a lot of information about the orientation
> > > issue, nothing seems to address this specifically.
> >
> > I don't fully understand what you did here. You flipped in three
> > directions,
> > which should have resulted in a flipping of the co-ordinate system from a
> > left-handed to a right-handed system (flipping twice would be just a
> > rotation).
> > In SPM99, flipping from a "radiologically" stored axial Analyze image, to
>
> a
>
> > "neurologically" oriented image, involves setting the initial estimates
> > to [0 0 0 0 0 0 -1 1 1 0 0 0]. This is a single flip in the x direction.
> >
> >
> > I don't really like the "neurological" and "radiological" notations.
>
> Left-
>
> > and
> > right-handed makes a lot more sense for MRI data, which can be acquired
> > in any orientation.
--
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
|