You said your EPI was a partial brain acquisition.
What parts of the brain are missing?
And you say that FNIRT overstretches, but what does the
the T1 to atlas registration look like? If this is not also
overstretching then it is likely to be small errors in the
EPI to T1 that are the problem. Please let us know if you
see any problems with the T1 to atlas registration.
All the best,
On 3 Jan 2012, at 12:57, Armin wrote:
> Dear Mark,
> Thank you so much for your answer!
> What you explain about registration in Melodic is totally correct. Melodic do all those step automatically. The EPI to the T1 (with using the linear registration) and the T1 to the Atlas (first with linear and then with non-linear registration). But, as you guessed in the nonlinear step we have overstretch. Although my EPI image doesn't cover all the brain but after FNIRT, it stretchs to the whole of the brain and even outside of the brain and I do not know how to overcome this problem. I would be so thankful if you can guide me to solve the problem.
> P.S. About the first part of registration, I should remind that I used both 6 ODF and 7 ODF to register the EPI to the T1 but the problem is not solve.