I found two different implementations of the code you adapted from Nolte et al. to calculate PSI and was wondering which one was correct. If I am not mistaken, the only difference between the two versions is in this line of code:
Version 1 : cs(:,:,freqi) = cs(:,:,freqi) + tempfftdat’*tempfftdat;
Version 2 : cs(:,:,freqi) = cs(:,:,freqi) + tempfftdat’*conj(tempfftdat);
I noticed that with the first version of the code, the connectivity value i to j is equal to - ( j to i ), whereas this does not occur with the second version of the code.