Hi Sven,
> - I had as idea to use DAT to DTI because TBSS
> non-linear registration might be more accurate for the deep
> structure. Do you have and comments on this?
It really depends on the quality of the contrast-to-noise ratio in your T1w images in the basal ganglia (don't count too much on the pallidum though - see fig 2 and fig 7 in Douaud et al., 2006). If there is a nice contrast, then fast4 within fslvbm will pick this up reliably...
On the other hand, you always have a very poor contrast between the lateral ventricles and the caudate nucleus in FA images, which might prove problematic in your case. On top of this, you'll certainly have distortions between the T1w and the dw which are likely to occur precisely in the caudate, ventral striatum and a bit in the thalamus above the sinuses.
> - Do you have a paper which I could use as reference?
No paper I'm afraid...
There is always the Reig et al., 2007 paper on problems of PET registration on the caudate, but as far as I remember, they deal with PET images with no corresponding MRI (which happens in quite a few clinical studies).
> - Could you please give me more information. In particular,
> did you use BET on T1? BET on DaT? Which parameters?
I was using Brainvisa and spm at the time so I'm not sure what's best here... Possibly DAT on betted T1w or even GM (haven't tried this)?
Cheers,
Gwenaelle
|