You can use a (hand-drawn) mask of the middle of the brain in FLIRT (the --inweight or --refweight options) so that the cost function is mostly driven by registration within those masks. The registration to standard space will be better in these masks at the expense of perhaps a worse registration at the edges of the brain.
You can also run dtifit on the dti data to get FA images that you can then register to the FMRIB58 brain (in $FSLDIR/data/standard) using FNIRT. You can do this easily with the tbss scripts (tbss_1 and tbss_2) which incorporate the appropriate call to FNIRT.
I forgot a 3rd possibility: you can draw the caudate masks by hand in fslview :)
Saad
On 29 Nov 2011, at 14:11, Li Yadi wrote:
> Thank you so much,Saad.
> I tried image derived from First on one subject and got worse mask.I think the steps i took are right.
> Can you give more details about the 2 alternatives you mentioned?
>
--
Saad Jbabdi
University of Oxford, FMRIB Centre
JR Hospital, Headington, OX3 9DU, UK
(+44)1865-222466 (fax 717)
www.fmrib.ox.ac.uk/~saad
|