Hi,
I want to use the random effect model on a group of 13 subjects (fMRI
study). My run has 8 blocks(7 scans per block) and my voxel
sizes are 2 x 2 x 4 after normalisation. I have been suggested that in
this situation, FWHM for smoothing should be between 8 and 16.
Another credible source, however, suggested that oversmoothing could
reduce my chance of finding clusters and recommanded to keep the
usual values (2 or 3 times the voxel sizes). What's the rational to
choose a proper FWHM for the random effect model? Even if I choose
between 8 and 16, what makes me decide if it is 8, 10, 12, 14 or 16?
Claude Breault Ph.D.
Research Center
Cognitive Neuroscience laboratory
Institut Universitaire de gériatrie de Montreal
Montreal, Quebec, Canada
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|