Hi Gang,

This is correct. I ran 3dREMLfit on all runs, and want to take the absolute value of dset a and dset b for each participant.

I ran the 3dcalc script you wrote for each pair of runs I need the absolute difference for, and this only was successful for 2 participants. For the other 6 participants, a 'fatal error' came up saying, "Multi-brick datasets don't match!". I am new to afni, and the preprocessing was completed and 3dREMLfit script was already written for each subject before I came onboard with this data. This is the 3dREMLfit script:

3dREMLfit -matrix X.xmat.1D -input pb04.3271456_20140102.mid1.r01.scale+tlrc.HEAD \
-fout -tout -Rbuck stats.3271456_20140102.mid1_REML -Rvar stats.3271456_20140102.mid1_REMLvar \
-Rfitts fitts.3271456_20140102.mid1_REML -Rerrts errts.3271456_20140102.mid1_REML -verb $*

I assumed that by the time this script was run, the voxels would be in a standard space. Did that not happen based on what you see here? Could the fatal error be because the matrices were not standardized?

Additionally, for the 2 participants where 3dcalc was successful, there was SO MUCH activation outside of the brain. Why is this happening? I know that a mask can be used so that we only see activation where the functional voxels are, but regardless of whether or not you use the mask, there shouldn't be SO much activation outside of the brain, right? I have not observed this with FSL or brainvoyager, and the amount of activation outside the brain really bothers me. Does this point to an error in preprocessing?

Thank you SO much for your help!!

Lindsay