Dear Esa,
> 1) I used The Full Monty.. model. After analysis I chose SVC and
> Search Volume... Image. Then I selected manually created mask image.
>
> 2) I tried explicit masking in The Full Monty... model using the same
> manually created mask image. Otherwise the Full Monty... model was the
> same as earlier.
>
> The statistics (Z-scores and p-values) were similar but not quite the
> same. Search volume and the number of resels were slightly different
> between SVC and explicit masking. The activated clusters were located in
> the same places and the number of voxels in each cluster were almost the
> same.
>
> What causes the slight differences between explicit masking and SVC?
The differences occur because explicit masking does not take account of the
smaller search volume in calculating corrected p values. Masking is just
like cutting a hole (or several holes!) in a piece of paper and peering
through it at the original SPM{t}. Clusters outside the mask cannot be seen;
clusters inside the mask can be seen, but their (corrected) signficance is
exactly the same as if no mask had been used. SVC, on the other hand, makes
a formal assessment of the new smaller search volume and calculates (new)
corrected p values on that basis.
The difference you observe between SVC and masking in corrected p values
will become greater as the masking image is made smaller. This is because
the search volume will become smaller and smaller for the SVC, but remain
identical for explicit masking.
> In my opinion, explicit masking can be used to visualize SVC
> results. Any comments on this?
SVC and explicit masking are different procedures, so are useful for
different questions. In general, SVC is useful when one can anatomically
restrict a prior hypothesis on the basis of a co-ordinate list or image mask
from an independent contrast. Masking may be useful in situations where it
is inappropriate to restrict the volume of search, but one wants to ask
questions about anatomical overlap between clusters activated in different
contrasts.
best wishes,
Geraint
--
Dr. Geraint Rees
Wellcome Advanced Fellow, Lecturer,
Division of Biology 139-74, Institute of Neurology,
California Institute of Technology, University College London,
Pasadena CA 91125 London WC1N 3BG
voice 626-395-2880 020-7833-7472
fax 626-796-8876 020-7813-1420
web http://www.klab.caltech.edu/~geraint
--
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|