Hi,
I am having difficulty using the FNIRT option in FEAT. I understand that
non-linear registration requires two anatomical/structural files. FNIRT
requires an initial affine registration based on a brain extracted
anatomical image as provided by FLIRT. In addition FNIRT also requires a
whole head image from which to derive the non-linear transformation. These
two requirements currently one of trigger two errors, either a non-linear
registration error (shrunken functional syndrome), or a linear registration
error (homunculous head projected onto MNI standard syndrome). My problem
stems from difficulty in providing both files. If I specify the brain
extracted image to be the "Main structural image", then a warning window
informs me that a "whole-head" image cannot be found, and that the
non-linear registration will be suboptimal. This warning appears despite the
presence of a single whole-head file with the suffix "_brain", (as per the
instruction in the warning message), in the same directory as the "betted"
anatomical. On the other hand, if an non-extracted anatomical image is
specified to be the "Main structural image", then the linear registration fails.
I have had success applying the non-linear warp files to the anatomicals
via "applywarp", although I have yet to successfully warp functional files.
I will appreciate any guidance, as I am eager to see what FNIRT can do!
Regards,
Kent Conover, Ph.D.
Concordia University,
Montreal, Quebec,
Canada
|