Hi,
It is really hard to tell what is happening from these jpgs.
Firstly, I would recommend performing the unwarping
via the FEAT GUI (in the prestats). This automatically
cleans up noisy voxels at the edge of the brain and does
other niceties.
You still need to get a correct rad_s fieldmap input though.
Presently, it doesn't look like you have the phase set correctly
as the rangeout_phase_rad_s and ring_phase_rad_s images
look full of wraps. Do these disappear after prelude is run?
That is, are the unwrapped_rad images smooth and without
sharp edges inside the brain? If there are wraps then either
the initial phase is not set correctly, or the unwrapping is failing,
which can happen with poor masking. In general we recommend
running BET on the magnitude image to generate a mask to use
with prelude.
As for the phase scaling, your processing only works if the initial
image ranges from -A to +A, in which case the output ends up
from -pi to +pi. Otherwise, if it is just 0 to +A then the output is
0 to +pi which is not correct (there needs to be a 2*pi range).
With respect to the corrected images, they show some streaks
which should be cleaned up by using a BET mask on the
magnitude image when you pass it into FEAT (or even into
the fugue command line if you prefer).
All the best,
Mark
|