In this thesis, the heritability of a number of features of brain functional connectivity and network behaviour was estimated, in order to assess their suitability as imaging endophenotypes. Phenotypes were chosen based on prior association with psychiatric disease in the literature, and current interest in the neuroimaging community. Heritability was estimated using a large MRI twin sample from the Queensland Twin Imaging Study.
The first study looks at intrinsic functional brain networks present at rest, in the absence of cognitive demand, and characterised them using graph theory, a mathematical formulation used to describe topological properties of complex networks. Such characteristics were found to be moderately to strongly heritable (h2=20-60%). The heritability estimates varied substantially with methodological choices, in particular the removal or inclusion of global signal.
Connectivity of functional brain networks during working memory performance were also examined. Dynamic causal modelling was employed to determine task-related changes in functional coupling between frontal and parietal regions. Changes in connectivity with task demand were observed for both forward and backwards connections, but the changes had low test-retest reliability (ICC≤0.3), and subsequently, twin correlations and heritability were non-significant (rMZ≤0.08, rDZ≤-0.05).
Finally the functional connectivity profile of the dorsolateral prefrontal cortex during working memory was probed. Functional connectivity with ipsilateral parietal cortex (h2=24%), contra-lateral dorsolateral prefrontal cortex (h2=36%), posterior cingulate cortex (h2=37%) and middle frontal cortex (h2=26%) were found to be heritable. However, the connections during baseline condition were also heritable, suggesting the heritability did not pertain to WM-specific connectivity, and connectivity with left hippocampus was not found to be heritable.