Extracting brain connectivity

Abstract This chapter shows that new methods for measuring effective connectivity allow us to characterize the interactions between brain regions which underlie the complex interactions among different processing stages of functional architectures. It reviews the basic concepts of effective connectivity in neuroimaging. The methods introduced to assess effective connectivity are multiple linear regression, covariance structural equation modelling and variable parameter regression. The first example demonstrates that non-linear interactions can be characterized using simple extensions of linear models, while in the second, structural equation modelling is introduced as a device that allows one to combine observed changes in cortical activity and anatomical models. Finally, the chapter concludes that the approach to neuroimaging data and regional interactions is an exciting endeavour, which is starting to attract more attention.