Dear Bernie,
>First, concerning the masking, if I understand correctly, I create new
>contrast images and re-estimate my model. But then, do I have to mask all of
>the subject images (1-12)? You wrote apply it to the first, the rest will be
>masked as well?
That's exactly right. SPM only retains voxels that have data for every
point in the data sequence, so you only have to mask the first.
>Also, the contrast image contains beta values, the masked
>only betas that survive some threshold, if I then re-estimate, and run an
>analysis again, it seems I have restricted my analyses to a smaller volume.
Exactly.
>Wouldn't small volume correction for my region in the initial analysis
>achieve similar (identical results).
Yes it would (if you used the mask to define the search volume). If you are
comfortable defining a restricted search volume, using a small volume
correction,
then you should do that. I suggested the masking option so that you could
threshold the SPM{T} (on masked data) at p=0.05 corrected and save the
surviving voxels. I am not sure that the small volume correction allows you
to save only significant voxels, but this is not necessary of you keep record
of what is significant and what is not.
>Second, I noticed to0 late that registration for the upcoming SPM course was
>open, (in fact, I checked regularly but never saw any form or something like
>that to apply for the course, but that's probably something I just missed).
>So my question is whether or not I could still join in for this course or
>not?
Oh. I am sorry but there is already a long waiting list. I apologise to
everyone
who cannot get onto these courses. The reason is we keep them small is to
enable
the practical special-interest groups, which means we can only cope with
about 64
people at a time.
With very best wishes,
Karl
|