 |
 |
When using clusterwise correction (3dClustSim, or the Cluster section of SPM results tables), both simultation based approaches and random field theory based approaches depend on the initial cluster threshold, not just the data smoothness. This is because what you are interested in is the probability of obtaining at least one cluster of k or more voxels. The size of a cluster clearly depends on the voxelwise threshold used to define that cluster (a more lenient initial threshold yields larger clusters, and so increased odds of finding a cluster of a given size by change, while a more strict threshold leads to smaller clusters and a correspondingly smaller change of finding a cluster of a given size by chance), so your clusterwise results will change depending on your initial voxelwise threshold. If you are looking at voxelwise p-values in the SPM results table, these should not change even if you change your threshold, because they do not depend on it. -Mike -- Mike Angstadt Research Area Specialist Lead / PANLab Lab Manager Department of Psychiatry / University of Michigan (734) 936-8229 -----Original Message----- From: SPM (Statistical Parametric Mapping) <[log in to unmask]> On Behalf Of Jingwen Jin Sent: Thursday, January 24, 2019 9:49 AM To: [log in to unmask] Subject: [SPM] SVC, 3dClustSim, and Initial threshold Dear SPM folks, Could someone explain to me why and how the initial p-threshold affects SVC and/or 3dClustSim correction? As for SVC, part of what I am confused about is that I thought SVC would calculate the Resels (resolution elements) within the (inclusive) masked area, and then correct the p-value based on random field theory. But then this would mean that the initial p-threshold would have nothing to do with the SVC-correction, which is clearly not the case in my experience. For instance, when I use the initial threshold of p = 0.001 uncorrected, and then do SVC correction, the peak significance values are the same as when I used initial p = 0.05 uncorrected. But when I use initial p = 0.05 FWE corrected, the same cluster would not even show up in both before and after SVC correction. As for 3DClustSim, I would like to know what are the differences in calculation compared to SVC in SPM? And pros and cons? Any answers and/or references are highly appreciated! Best, Frances ********************************************************** Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues