Dear fellow ynic-users,
You may notice some difficulty removing non-brain tissue from structural MR
images obtained at YNiC when using the brain extraction tool (bet - a part
of FSL) with its default parameters. Symptoms are that tissue around the
brainstem, mouth and throat is not removed. This might adversely affect
coregistration.
One workaround I know some people have used is to coregister the
"unstripped" brain to the "unstripped" reference volume (which seems
to work
quite well). However, the FSL help strongly recommends removing non-brain
tissue, so this might not be the best solution.
Bet uses an estimate of the centre of the image to begin the process of
finding the inner and outer scalp and brain boundaries. The default estimate
doesn't seem to be right for out structural scans. I have had more success
using the following from the command line:
bet myinfileroot myoutfileroot -c 87 108 165
Where you replace myinfileroot and myoutfileroot with the names of your
input and output files (without .hdr or .img on the end).
The coordinates 87 108 165 may not be quite right for the centre of your
image but the bet algorithm seems to work much better with these values.
Best wishes,
Tom Hartley
===================================
Dr Tom Hartley
Dept. of Psychology,
University of York
Heslington,
YO10 5DD
UK.
t.hartley(a)psychology.york.ac.uk
http://www-users.york.ac.uk/~th512
tel: +44 (0)1904 432903