Are you providing FNIRT with the non-brain-extracted
versions of the image and a good affine initialisation?
If so, and you cannot see any other problems with what
you are doing then feel free to upload the dataset to:
and make sure you send us the reference number.
All the best,
On 30 Aug 2010, at 18:00, Elizabeth Losin wrote:
> Hello all,
> I am using FNIRT to perform nonlinear registration on a sample of
> 19 participants. My registration process uses an intermediate in-
> plane high-resolution scan to aid in registration of the EPI images
> to the standard. The MPRAGE nonlinearly registered into standard
> space for one participant is distorted at the top of the brain.
> The brain looks caved in on one side. These distortions are not
> present when linear registration (12 parameter affine) is used
> instead. Currently, I am using the FNIRT defaults suggested in the
> FNIRT documentation for a functional study. Are there any
> strategies for fixing this distortion within FNIRT? Any suggestions
> would be greatly appreciated. Thanks.