Dear John,
Is it possible that the second approach smoothed only within the brain mask, while the first approach smoothed the entire 3D volume? If so, for the first approach a lot of the signal of interest might have been lost on brain edges. This would explain why the p-values for the second smoothing were lower (if indeed there is a real difference between patients and controls). I guess it is better to smooth only within the brain mask (-> second approach).
Best,
Wiktor Olszowy
|