I've never tested age on the entire IXI dataset, so I'm not entirely
sure. What were the other columns in the design matrix? There are
data from a number of scanners in the dataset, so you'll need to
include scanner as a confound. Did you do any type of global
correction? If you used proportional scaling to correct for whole
brain volume or total GM volume, then it's possible that you may see
some apparent increases with age.
Note also that darkening of white matter could give it an intensity
similar to grey matter (and hence apparently more GM in a VBM
analysis). Partial volume effects can also do this (so its common to
see an apparent GM increase around the ventricles, which arises from a
larger WM/CSF boundary with age).
Other than that, I'm not entirely sure.
Best regards,
-John
On 15 July 2011 22:12, Josselin <[log in to unmask]> wrote:
> Dear all,
>
> I am training a little on the IXI-T1 dataset for VBM. I studied the influence of age on GM density, following the principles of VBM with DARTEL by J. Ashburner (www.fil.ion.ucl.ac.uk/~john/misc/VBMclass10.pdf).
>
> When I look to voxels where GM density increases with age, I got a strange result: many cortical areas emerged as having their GM density increasing with age (I expected no brain regions to pop up).
>
> I find this very intriguing. Do you think I have made an error ? Or is it plausible ?
>
> Thanks !
> Josselin Houenou
|