Dear Jay,
> I've discovered something of interest...
>
> I mentioned that I used fslroi to crop the images prior to bias
> correcting
> and bet-ing. I had asked if the differing image sizes would be an
> issue for
> fslvbm, and Gwenaelle replied that it would not.
>
> Since you seemed surprised that my subject volumes looked quite
> different
> from each other in the GM_mod_merg file, I have run an experiment
> using one
> subject only, a copy of that subject, and a version of that subject
> cropped differently.
> The original and copy look identical after processing through the
> fslvbm
> scripts, however the newly cropped volume is visibly different from
> the other 2!
>
> Clearly the cropping introduces some alteration, and the widely
> different cropping
> parameters for my subjects my seems have contributed to the visible
> differences. I
> used the non-linear version of registration btw...would affine been
> a better choice?
I am not surprised you find a difference, but I would have thought
that the results you get with cropping would be the better ones. The
main way in which cropping will affect your results (apart from
reducing processing time and disk space) is by allowing BET to do a
better job.
Maybe by now it is better if you uploaded the mod_merge file so we can
have a look at it.
Jesper
|