Dear Rick,

I just tried it again with even weirder results.

I tried it with the input file masked and using the -mask option. This time it did not blur within the mask at all. The input and output datasets are identical.

Again, when I blur the entire dataset with the mask option all voxels are blurred, inside and outside of the mask.

Just to make sure, the mask is a dataset where it is value of "1" for the region of interest and zero elsewhere. Is that right?

Below is the output from the screen. It sees the mask!

3dBlurToFWHM -input decon_HitFACRTwoLevelsMatchingforConjunction_lhip_only+tlrc -FWHM 4 -mask lhip2mm+tlrc -prefix decon_HitFACRTwoLevelsMatchingforConjunction_real4mmblur_lhip_only
++ 3dBlurToFWHM: AFNI version=AFNI_2011_12_21_1014 (Sep 18 2014) [64-bit]
++ Number of voxels in mask = 487
++ Using input dataset as blurmaster
++ Max number iterations set to 133
++ Using default neighborhood 'SPHERE(-4.0)'
++ Neighborhood comprises 257 voxels
++ Using blurmaster sub-bricks [0..18(1)]
++ -- Iteration #1: 3D FWHMx=5.2771 FWHMy=3.7361 FWHMz=3.8616 cbrt()=4.2383
++ ** Passes 3D threshold cbrt(FWHMx*FWHMy*FWHMz) ==> done!
++ Output dataset ./decon_HitFACRTwoLevelsMatchingforConjunction_real4mmblur_lhip_only+tlrc.BRIK
++ Checking results by running command below:
+ /usr/local/afni/bin/3dFWHMx -arith -mask ./lhip2mm+tlrc.BRIK ./decon_HitFACRTwoLevelsMatchingforConjunction_real4mmblur_lhip_only+tlrc.BRIK
++ 3dFWHMx: AFNI version=AFNI_2011_12_21_1014 (Sep 18 2014) [64-bit]
++ Authored by: The Bob
++ Number of voxels in mask = 487
5.53484 4.13402 4.30545


Christine