Hello,
I'm relatively new to FSL and I've been using fnirt for registration
to the MNI template. My workflow is very standard: a linear
registration of subjects' bold images to T1 structurals, then the
nonlinear warp from T1 to MNI.
My problem is that fnirt invariably smudges the edge of the brain,
"pulling" brain voxels towards the skull. This is a problem because
much of the bold signal of interest is in the cortex and is therefore
distorted by this warp. I'm currently registering 2x2x2mm T1 images to
the 2mm MNI brain, using all the default parameters. (with a lesion
mask as these are stroke patients, but I also see the smudging in
controls). Is there anything that can be done to reduce this
distortion?
thanks!
-bgroisser
|