| Summary: | An increasingly important concept in psychiatric neuroimaging is that of brai nconnectivity. Dynamic Causal Modelling (DCM) has been successfully usedto infer how spatially remote areas of the brain integrate to form functionalnetworks. A potential disadvantage to DCM is the need to predefine a modelbased on a hypothesis about the underlying connectivity. This requirementsmeans the results are dependent on the assumptions about model structure,and important features of the underlying network may be ignored. Here wepresent a method for identifying the model structure in a way t hat discardsthe a priori knowledge that is typically used to constrain model space. Thisallows DCM to be used in a more data-driven way, and allows the optimalmodel within a network of nodes to be identified. The thesis consists of 3studies that together provide a generic framework for a novel approach toDCM and validation that it works, and offers a significant computationaladvantage to traditional DCM.The first study demonstrates that the connectivity within a system of brainregions can be ascertained from inferring the connectivity within smallersystems, which consist of regions taken from the entire system. By analysingthe data in this fashion, we can effectively explore the entire networkstructure space, but estimate a much smaller number of models than wouldbe typical. The second study applies the method to a multicentre dataset andshows that Bayesian Model Selection (BMS) results are reproducible atdifferent centres and across different sessions. The findings show that DCMis robust enough to be used in multicentre studies and that our exploratoryapproach is just as effective as traditional approaches to DCM. The thirdstudy applies the method to a standard psychiatric imaging dataset; animplicit emotional processing face recognition task performed by patientswith major depressive disorder (MDD) vs healthy controls (HC). The MDDpatients perform a follow up scan having being treated with theantidepressant citalopram. The study shows that the developed method canbe used to identify the optimal model structure in order to make inferenceson effective connectivity parameters, and identify differences between patientand control groups, and before and after treatment.
|
|---|
