Unless you have exactly the same amount of atrophy at every single
point on the boundary, you expect atrophy to induce a change in
shape and not just a global change in size. If the atrophy is
then it is going to be a strong combination of both, but if it is quite
local (as in "local atrophy") then it is mainly represented by local
changes in shape rather than a change in size. So I would generally
expect both shape and size to contribute to atrophy, but given the
general size change within the normal population, it can sometimes
be more helpful to normalise for the global size change and just
look at changes in shape (local atrophy) as it can be more sensitive.
However, the best thing to do will depend on the particular study in
I hope this helps clarify things.
All the best,
On 20 Aug 2009, at 15:48, j janssen wrote:
> Assuming you're using native space surfaces, then the removal of scale
> allows you to look at "shape" changes. In other words, local atrophy
> within the structure and omitting any isotropic global scaling.
> i think (thought) that atrophy is a decrease in size, it is not a
> change in shape. 2 objects can have equal sizes but different
> shapes. thus, i think that findings from a straightforward size-and-
> shape analysis cannot be interpreted as atrophy.
> > Hi,
> > in Brian's thesis it says (page 177-178):
> > "In neuroanatomy, variation in the global scale of a structure may
> be of
> > interest as it may correspond to variations in grey matter volume
> due to
> > pathology. Thus strictly speaking we are usually interested in
> > investigating
> > the variation in size-and-shape of the deep grey structures.."
> > on the FIRST website:
> > http://www.fmrib.ox.ac.uk/fsl/first/index.html
> > it says:
> > *"--useScale* : Used in conjunction with *--useRigidAlign*, it
> will remove
> > global scalings when aligning meshes."
> > "Also note that the --useScale option will not be used. Without the
> > --useScale option, changes in both local shape and size can be
> found in
> > shape analysis. This type of finding can be interpreted, for
> example, as
> > local atrophy. With the --useScale option, overall changes in size
> > lost."
> > i'm not sure i understand. No --useScale means that my metric does
> > separate size from shape, i.e. it is a size-and-shape analysis. i
> do not
> > see
> > how my finding can then be interpreted as local atrophy (size).
> > including --useScale or some kind of size variable in my
> statistical model
> > assures, at least, that my metric is interpretable as shape, i.e.
> it is a
> > shape analysis.
> > -joost