Hi,
I'm wondering whether there is a possibility to test whether the assumptions required by random gaussian field theory are fullfilled with a special FWHM setting ?
Best wishes
Stefanie
Hi,
essentially this depends on the anatomical size of the expected effect. Using a 8 mm FWHM kernel will make difficult to detect small activations smaller than the FWHM since the response can be effectively killed by the blurring. On the other hand if you use very small FWHM your data may not be smooth enough to fulfill the assumptions required by random gaussian field theory when you correct for multiple comparisons. The standard FWHM is most likely to reflect a balance between these two extremes.
Bw,
Enzo
On Mon, Mar 26, 2012 at 2:02 AM, Sam Smith
<[log in to unmask]> wrote:
Dear All,
When I read neuroimaging articles, I encountered with a doubt reg.smoothing.
smoothing is done to minimize noise and residual differences in gyral anatomy with a Guassian kernel.
If i am correct, FWHM ranges between 8-12mm.
On what basis we select this value.
CAn I use any value in between this range??
--
Regards,
Sam Smith