OK, I'll take it.
It's not clear to me what is going wrong or exactly what you've done
with your registrations at the first level. However, I would not
recommend
using the FIRST registrations either way as these are just for
structural
(highres) to the MNI (standard) but are only simple linear
transforms. It is
much more likely that you'll get a better registration using non-
linear registration
(FNIRT) instead. As this seems to be the only thing likely to have
caused
a problem then I'd avoid it (and I don't know exactly what the problem
would
be, but if it either created much higher resolution images or had some
complete registration failures, then it is possible that you might see
such a
slow-down - although I'd still be a bit surprised even then).
Anyway, my advice is run your analysis with the standard registration
protocol as it is less likely to be problematic and also likely to
give you
better results. This will hopefully run in a few hours, as you say it
has
done before. You will probably want to re-generate your hippocampal
mask though by using applywarp with the FNIRT results to get each one
into standard space and then averaging and binarising that mask (as this
is what I assume you've done originally to create your hippocampal
mask).
All the best,
Mark
On 20 Aug 2010, at 14:51, Carlos Faraco wrote:
> Any takers please? R&R deadline is coming up soon.
>
> Thanks,
>
> Carlos
>
|