Hi Rick,
Thanks for the help. I was hesitant using the -demean function due to the fact that the mean changes from region to region across the brain mask. I think that if I create a mask of each separate region and then apply the 3ddot with -demean option this should probably address those concerns.
Could another option be to actually create a voxelwise mean map (ie 3dLocalStat) of each parameter and then use 3dcalc to substract the respective mean maps from the files. Then if I use 3ddot WITHOUT the demean option, this should essentially do a pearson's correlation as well while taking into account the mean more precisely?
Thanks,
Ajay