Is there a way I can select which clusters are rendered in the post-stats, by cluster size?
We tried running our 3rd level analysis with cluster thresholding in the post processing stage. Depending on the commonly used (1.75 or 2.3) levels of thresholding, we get some activation in the A<B contrast and no activation in the A>B contrast, or some activation in A>B and one huge cluster in the A<B contrast.
We therefore tried running a 3rd level analysis uncorrected with a p=0.005 (Z=2.6) and then a cluster analysis on the stat-files. By considering only the clusters with a physiological smoothed size (i.e. 216 voxels because of 1 mm^3 voxels and a pre-processing smoothing of 6 mm), we get significant activation in both contrasts in clusters. But the cluster program only outputs in text, which makes getting an overview of the data a very painstaking manual process.
|