Hi Emily,
The gmean.errts.unit.1D vector is created on the way
to computing GCOR (the global correlation: the average
correlation of every pair of (brain) voxel time series).
It is a single number measuring global connectivity in
a time series.
That is a slow and CPU intensive computation. But if
the time series are demeaned and scaled to unit length
(sqrt(sum squares) = 1), the correlations are just dot
products, and the average of all pairs of dot products
is the average dot product at each voxel with the average
time series, which is simply the dot product of that
average with itself. And that is just the sum of squares
of gmean.errts.unit.1D. It becomes fast and easy.
Regarding global signal regression, that can be done
with "afni_proc.py -regress_ROI brain".
However, please try not to use GSR, except perhaps
to show that it can produce strange results. Global
signal regression alters the correlation matrix that you
are otherwise trying to measure, and in an unknown way.
- rick
Edited 1 time(s). Last edit at 07/18/2014 09:24AM by rick reynolds.