> I have 2 questions regarding snpm.
> 1. Is the statistic method in snpm and randomise under FSL same ?
I'm not familiar with RANDOMISE in FSL, so I let others answer the first
question.
> 2. I plan to use snpm3 to perform statistic analysis. However, when I
use
> smooth function in setup snpm, I always get out of memory error. Could
I
> smooth my image firstly befor I start to setup snpm, than sent these
> images into snpm setup and set FWHM smooth to "zero" under snpm ? Is
it
> reasnonable ?
I presume you are referring to the variance smoothing in SnPM. The
variance smoothing in SnPM only smoothes the variance image, which is
the denominator of your T-statistic image. It's different from smoothing
your images first and doing an SnPM analysis without a variance
smoothing. If you smooth your images beforehand, you are implicitly
smoothing both numerator and denominator of your T-image.
-Satoru
|