New methods to estimate functional relationships between ROIs
Peter Molenaar
Department of Human Development and Family Studies
Penn State University


Abstract:

The currently popular approaches to estimate ROI connectivity maps include structural equation models (SEM), vector autoregression (VAR) and dynamic causal modeling (DCM). Each of these approaches has limitations. For instance, SEM is confined to contemporaneous relationships and neglects external input; VAR is confined to lagged relationships and often neglects external input; DCM is based on an underlying dynamic model which is deterministic. In addition, these approaches cannot be applied in a purely exploratory mode. Our new approach is based on state-space modeling, involving a combination of SEM, VAR, and linear/bilinear effects of external inputs. The state-space model is fitted to the data by means of unconstrained or arbitrarily constrained automatic search based on repeated Lagrange multiplier testing. Simple applications to simulated and empirical data will be presented.