> how we can choose good FWHM for smoothing?
It depends on what you want to do.
>
> for example in VBM, if one want to make template, must smooth it with FWHM
> = 8mm.
Do you mean template, or tissue probability maps. Templates should be
smoothed by about the same amount as the default setting for the source image
smoothness.
SPM2 or SPM5? For SPM5, no smoothing is usually recommended for tissue
probability maps because some nonlinear warping is included within the
segmentation model.
>
> is this emprical choise?
Really it should be an empirical choice. The data in the apriori subdirectory
are smoothed in order to make the tissue probability maps a little more
general for brain images that deviate a long way from those that were used
for generating the mean.
>
> and how one can change it?
> is it relation between FWHM and voxel dimentions?
I guess that an additional parameter could be included in the segmentation
model in order to select the optimal smoothness. This is unlikely to happen
very soon though.
Best regards,
-John
|