It depends. I believe the general rule of thumb is 2-3 times your voxel size when you're using GRF to threshold. The main thing to consider, however, is the expected size of your activations (cf. matched filter theorem). For example, if you're predicting activation in a small region such as the ventral striatum, you probably don't want to use a large smoothing kernel (e.g., 10mm). Instead, something smaller -- like 4 to 6mm -- might be more appropriate.
Hope this helps.
David
--
David V. Smith
Graduate Student, Huettel Lab
Department of Psychology and Neuroscience
Duke University
Durham, NC 27708
On Nov 27, 2009, at 12:15 PM, MRI student wrote:
> Hi --
>
> What range of spatial smoothing kernels do you consider appropriate? I know that 5 mm is your default. Is it possible to use a smaller kernel such as 3mm or 4mm?
>
> Thank you.
|