On Mon, 15 Mar 1999, carol docherty wrote:
>
> Hi,
>
> I'm currently analysing brain volumetric changes in
> patients' with Alzheimers. I have two 3D MR datasets
> from the same subject taken six months apart and I'm
> looking for changes in brain volume at the voxel level
> (looking for voxels whose intensities have changed by a
> certain amount).
>
> Prior to any analysis, I have realigned the two images
> together and then carried out a 6-parameter
> normalisation on them. However, what I'm finding is
> that the number of voxels within each volume changes
> drastically following realignment, and not by the same
> amount either ie following realignment, the actual
> difference in volume between both datasets increases
> significantly. I have tried to make the images as
> similar as possible before realignment. Why should the
> brain volume change following realignment?? I
> understand that it will change at the normalisation
> stage but surely by the same amount for both datasets.
>
> Thanks in advance,
>
> Carol.
> ----------------------
Hi Carol,
If you are using the realign option in SPM96 its' default is to have image
masking on. It's very likely that this is causing you to lose voxels on
the realigned images. You can turn masking off in the defaults topion in
SPM. However, it may not be sensible to include voxels from space where
there was no image prior to realignment. You may have to look closely at
the dfata with and without masking to see what's actually happening.
Regards - Mike
--
_____________________________________________________________________
Mike Glabus PhD, Principal Physicist
MRC Brain Metabolism Unit Tel Intl: 44-(0)131 537 6818/6282
Royal Edinburgh Hospital FAX Intl: 44-(0)131 537 6110
Morningside Park Pager : 5236, via 0131 537 6000
Edinburgh, EH10 5HF, SCOTLAND
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|