afni.nimh.nih.gov

Hi Matt,

I ran a quick test on this, comparing the "3dTcorrMap -Mean"
output with the method used in afni_proc.py and the method
used in @compute_gcor, and they do match, subject to
truncation errors.

The voxel-wise ratio (a-b)/(a+b) between the 3dTcorrMap
output and the afni_proc.py output is less than 0.01 (abs)
everywhere. So the results seem very consistent.

Maybe I should add an option to @compute_gcor to save
the correlation volume, too.

- rick