Same-same or different? Brain network characteristics during rest and tasks in major depressive disorder

For centuries, understanding depression has been a central scientific endeavour across numerous scientific disciplines. Contemporarily, a biomedical perspective has won favour and is dominated by biomarker research, which seeks to identify objective pathophysiological features of major depressive disorder (MDD). One such approach utilises functional magnetic resonance imaging (fMRI) to measure brain function, two key methodologies being resting-state (rsfMRI) and task-based (tbfMRI) investigations, ordinarily kept separate in analyses and empirical articles. Psychologically, MDD is frequently conceptualised as a disorder of rigidity or abnormal reactivity, which may manifest as stable or aberrantly fluctuating patterns of brain activity, respectively. Recognising this, this dissertation explores the potential of combining rsfMRI and tbfMRI data to identify context-specific manifestations of MDD. To this end, MDD is presented epidemiologically, clinically, and aetiologically, followed by a theoretical and methodological introduction to fMRI, functional connectivity (FC), and independent component analysis (ICA). Then, literature on MDD abnormalities across six key FC networks and the concept of task-dependent brain states is presented. Next, utilising a dataset of MDD and controls scanned across three conditions (resting-state, emotional faces, reward), this thesis investigates four research questions regarding 1) the reproducibility of a novel network atlas, 2) the effect of scanning conditions on network metrics, 3) whether ICA-based biomarkers are discernible in the sample, and 4) whether pooling rsfMRI and tbfMRI is favourable when assessing network metrics in MDD. Analyses achieved overall replicability of the atlas, found widespread effects of conditions on FC, and identified previously reported and novel putative biomarkers, of which a range were condition-specific and suggestive of abnormal reactivity. The study supports the pooling of rsfMRI and tbfMRI data when probing MDD FC, but affords replication attempts and tailored multiple comparison correction strategies. Finally, this thesis discusses the inferential, translational, and ethical cautions of its contribution to the field, and encourages future investigators to utilise richer datasets in their investigations and transparency in their reporting of MDD fMRI biomarker research.