is there a standard protocal to use alphasim

Submitted by chunhuichen on Thu, 07/24/2014 - 15:38
Hi experts,
 
I just read this paper (Bennett,2009,SCAN, The principled control of false positives in neuroimaging ), in which the auther wrote:
Another method that is often incorrectly used is
the AlphaSim tool included in AFNI (http://afni.nimh.nih
.gov/afni/). For effective false positive control, AlphaSim
requires an estimate of the spatial correlation across voxels
be modeled using the program 3dFWHM. Many researchers
simply input the amount of Gaussian smoothing that was
applied during preprocessing, leading to incorrect clustering
thresholds as output. Errors during estimation of the spatial
smoothness can also lead to incorrect values.
I had used the preprocessing smoothness in REST AlphaSim so am I doing wrong?
 
I guess the Estimate button in REST AlphaSim is doing this correction ,right? Someone recommond estimate smoothness with 3dFWHM, which was in the AFNI package, did REST AlphaSim used the same algorithm as 3dFWHM to estimate smoothness?
From REST/Viewer/Misc/Correction threshold by alphasim, I can also get alphasim correction results, does this function estimated smoothness and how?
 
I searched on this Rest Forum,
this early post says just use the preprocessing FWHM core is OK, not necessary to use the estimated smoothness http://restfmri.net/forum/node/1937 
this other poster says, one can use SPMT image or residaul image to estimae, or use smoothness stored in SPM.mat (Probably in SPM.xVol.FWHM), but they are differ from each other  http://restfmri.net/forum/node/1857 
 
So is there a standard protocal to use alphasim? or is there some reference to argue against Bennett,2009,SCAN that using preprocessing smoothness is also OK?
 
Thank you for all your help and best wishes!