Dear SPMers, I have a question about data extraction in VBM analysis. My results suggest that there are some statistically-significant differences of GM concentration, but not volume, between my study groups in certain areas of the brain. If I wanted to correlate the GM density of these areas with some continuous variables (metabolites, scores, etc.) across the entire study population, I presume the more straightforward way of doing that would be to extract the individual sterological data (possibly corrected for the effects of interest) and carry out a Pearson's (or Spearman's) correlation analysis (as described in Mechelli et al, Nature 2004,431:757). But, when it comes to data extraction through the VOI button in SPM, I can choose to go either for a single voxel (sphere radius 0 mm) or for the entire cluster. Which choice is more appropriate: single- voxel or cluster extraction? Why? In the paper I cited above, as well as in several other publications, I could not find an accurate description of the procedure used. I would deeply appreciate any hints, in layman's words. Thank you very much in advance. Best regards, Nicola Pannacciulli