afni.nimh.nih.gov

> I would like to know the correlation at cluster level.

The problem is that there is no legitimate way to "average" the correlation coefficients across voxels within an ROI. You may want to consider averaging beta values at cluster level instead.

Gang