---------- Forwarded message ----------
From:
Paul Zhutovsky <[log in to unmask]>Date: Fri, Feb 3, 2017 at 5:32 PM
Subject: Registration with FNIRT question
To:
[log in to unmask]Dear FSL experts,
I have a question about my registration pipeline. Image summarizing the registration can be found
here (dropbox-folder).
I'm trying to register EPI data (TR=2s, 3mm isotropic) to MNI space using the standard FSL way as implemented through FEAT.
That
means I use the BBR cost-function to co-register my functional (middle
image) to my structural image and I calculate the final structural to
MNI registration by first using FLIRT (12 dof, affine transformation)
followed by FNIRT (10mm warp).
As you can see in the linked
image the result of the registration procedure for the
functional to MNI looks rather odd (there is a 'horn' of matter outside
the MNI's brain shape). This one data set performed especially bad but
some of the other data sets look comparably (bad) in terms of their
final registration.
Do you maybe have some ideas/insights why this might be happening and how to fix it?
Cheers,
Paul