hi Cyril
> Because FWE is a height threshold technique it is like FDR bonferonni etc ..
> however instead of voxels it uses resels and resel count depends on the
> smoothness, right? so I always thought that FWE does take into account
> (incidentally) the spatial distribution of the signal .. am i completely
> wrong?
I don't think you're completely wrong, but I don't know if that's the
best way to think about FWE. The resel size gives an indication of
the overall smoothness of the whole image---which includes signal +
noise. I.e., if you generated random noise and smoothed it, you would
still have fewer resels than voxels due to the smoothing. If your
signal is a relatively small proportion of the total image, it seems
safe to assume that most of the smoothness reflected in the resel
count is NOT due to your signal.
(If I'm wrong, hopefully someone more knowledgeable will be kind
enough to comment!)
Jonathan
|