Question about the GRF correction

Dear experts,

I use the GRF  correction in DPABI. I would like to confirm the setup in p-voxel and p-cluster. Can we use the p-value 0.01 for voxel-level and 0.05 for cluster-level correction now? Someone suggested that we should use the p-value 0.001 for voxel-level correction. If so, which value we could used for cluster-level correction? 

Many thanks in advance!

After Eklund's paper, I guess you have to use p<0.001 or permutation test in DPABI.