Hi all,
> And here is the Rorshach ink blot problem of functional imaging.
>
>When I first started giving small talks about imaging, I used to put
>up an SPM map generated from random numbers and ask the audience to
>tell me what experiment it had come from - the answers were very
>convincing.
>
>
Following the morning discussion let me add that Matthew's last year
comments on the use of unthresholded maps are of particular interest.
Whatever the correction we use, it's a good idea to look at raw t-maps -
on this topic, the 'ring' analysis proposed by Spiridon et al. last
month in HBM seems interesting: it does not control for multiple
comparisons (so we still need to do it) but shows that the blobs of
interest are not 'artifacts' of the threshold procedure.
Best
Cyril
|