Hello FSL Experts,
I've been playing around with the versions of FIRST pre and post 4.1.4 (when major changes were integrated) and looking at how it affects the means and distributions of the subcortical brain volumes in our group of subjects. One thing I noticed was that the newer version appears to be more conservative in it's estimation, which appears to restrict the range of the distribution across most subcortical brain regions. For example, the right amygdala the new version (after intracranial volume adjustment and transformed into cm3) has a range of .75 - 1.75 cm3 whereas the old version produced a range of 1.5 - 3 cm3. Moreover, the volume estimations for a lot of the structures are significantly different. I inspected all of the structures and the program appears to have done a decent job, but the fact that the ranges differ so significantly is concerning. I was wondering if someone could please speak to this and explain which version should be used and why.
Also, in a few cases the estimation was not completely accurate so I was wondering what is the best way to manually intervene on these participants to see if the volume estimation can be improved? Is it better to do a non-linear registration to the MNI152 or play around with the boundary correction? If playing around with the boundary correction is a good method are their rules to follow in regards to choosing the best strategy? It wasn't entirely clear to me based on the website.
Any advice is truly apprecitated.
Cheers,
Dana
>>> and then apply FIRST to the output
|