Many people seem to use simply linear registration for normalization.
Although , most of the things I've looked at use either T1 or T2 images.
I'm not sure how things will work out if you normalize directly the DTI
I don't have the exact parameters with me on what I've used T1 image
send along separately to you a pdf of a poster from the 2007 HBM
if you'd like.
I would try decreasing the number of nonlinear iterations, and increasing
the nonlinear regularization significantly.
I also like to use a mask since so much of the image is typically non-brain.
Hui, Edward SK wrote:
> Dear SPM experts,
> I have been trying to normalize several rat brain DTI data together
> using SPM5 with the following estimation parameters:
> nonlinear regularization: 40
> nonlinear iterations: 54
> nonlinear frequency cutoff: 25
> After the normalization, certain parts of the brain are not registered
> nicely while certain parts are acceptable (visually). So I am
> wondering if anyone could kindly tell me what sort of parameters I
> should change to improve the normalization?
> Any help would be greatly appreciated.
> Many thanks,