Which mask to use for smoothness estimation (FWHMx,y,z)?

Dear Chaogan,

My study is to use voxel-based rs-FC measure to perform statistical analysis. That rs-FC measure was generated within a GM mask, so that FC was calculated between any two voxels within the GM.

After statistical analysis,  however, I consider that the whole brain mask, instead of the GM mask, should be used for smoothness estimation (FWHMx,y,z) with 4D residuals.

In your previous reply to a question for rs-fMRI study, you also suggested using a whole brain mask generated by DPABI-Quality Control - Generate Group Masks for smoothness estimation (http://rfmri.org/content/questions-related-rest-alphasim). 

So in my study, is it correct to use the whole brain mask generated by DPABI-QC to perform smoothness estimation? Many thanks!

Best regards,

Haobo