[MINC-users] minctracc -measure output
elo@neurorx.com
elo at neurorx.com
Wed Aug 31 09:00:08 EDT 2005
Hello,
We are looking for ways to assess the quality of intermodal registration.
For example, the quality of alignment between T1 and T2 images, or T1 and
PD images. The purpose is to be able to automatically detect whether or
not our alignment algorithm has succeeded or failed.
One way of doing this is by using minctracc with the -measure option.
However this returns the zscore, xcorr, var_ratio, and ssc indices. But
it does not return the value of the mi (mutual information) coefficient.
Since 'mi' seems to be the most effective objective function for
intermodal alignment, we'd like to know it's optimized value.
Was mi omitted from the -measure output because of oversight (since it is
a recent index used in mritoself)? How might one evaluate the mi function
between two images, short of rerunning mritoself and looking at the
output? A more general question, does anyone know of a reliable way to
assess intermodal registration quality (i.e. success/failure)?
Thanks very much,
Ernest Lo
NeuroRX Research
Montreal, Canada
More information about the MINC-users
mailing list