> I checked the result of fnirt registration after using the conf
> file. Its result is
> better than the previous fnirt that I was using, but still is far
> from what
> expected. Let me explain the problem in more detail. I have two MRI
> with their manual labels (subcortical labels). The volume of
> structures of both of them is around 132,000 mm^3. When I register
> two images using fnirt with --refmask and --inmask (Registered
> mask to these MRI images) and then propagate the labels using
> the resulting subcortical structure has the volume around
> 167,000mm^3 in the
> best case using config file you recommended me. It shows that the
> subcortical structures after registration is expanded around 20%. I
> don't have
> this problem when I use fnirt without --refmask or --inmask. Any
> comment is
> highly appreciated.
just to add to MJ's reply.
MJ is right that the potential advantages of using masking (better
alignment since it is allowed to ignore structures outside the mask)
are less than what is the case for e.g. affine registration. It should
nevertheless work and at least not lead to worse results. It would
therefore be good to have a look at your data to see what is going on.
So if you could please upload your data to http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
and then send me the reference number? Upload all the relevant data
(ref and in scans, all masks you are using and the configuration file
you are using along with the _exact_ command line).