I have been working on a program that estimates the connection strengths within a network in the brain. Unlike the pure data-driven approach adopted in BrainVoyager in which the analysis focuses only on the connection of a voxel with a seed region in a bivariate fashion similar to simple correlation or context-dependent correlation (aka PPI) analysis, I decide to take the route of multivariate analysis, similar to structural equation modeling (SEM), by considering a number of pre-selected regions. But unlike SEM, the primary causality analysis with the program in AFNI is done with all the regions' time series at individual subject level.
I pretty much have a primitive version of the program, but need to test its reliability with more data. If anyone is interested and willing to ride through the bumpy road of testing, please let me (gangchen@mail.nih.gov) know.
Gang