Tomas,
> May I have a question? From your paper describing your method
correcting
> for nonstationarity of smoothness I gathered that even present
permutation
> approach as implemented in snpm3 is quite robust for data with spatial
> pattern of smoothness. Or am I wrong? I, clearly, am not a math
expert, so
> I may not understand it correctly.
Yes, you are correct. Even without the non-stationarity correction,
p-values from SnPM are quite accurate. But I must point out that cluster
test in SnPM is more sensitive to clusters in smooth areas than rough
areas. In other words, sensitivity is not uniform within the brain.
-Satoru
|