Functional Magnetic Resonance Imaging of Goal Maintenance in Schizophrenia: Activation, Functional Connectivity, and Reliability

Abstract

Cognitive deficits are some of the most debilitating and difficult to treat symptoms of schizophrenia. Goal maintenance is a facet of cognitive control that has been shown to be impaired in schizophrenia patients as well as their unaffected first-degree relatives. Previous fMRI activation studies found less activation in dorsolateral prefrontal cortex (dlPFC) in schizophrenia patients compared with healthy controls during the completion of a goal maintenance task. This dissertation consisted of a series of studies employing a large, multisite retest dataset of schizophrenia patients and healthy control subjects. These studies sought to replicate previous activation findings using a newer goal maintenance task, to use group independent component analysis (ICA) to determine if schizophrenia patients also exhibited dysfunctional functional connectivity or functional network connectivity (FNC) compared with healthy controls during the performance of that task, and to evaluate the test-retest reliability of each of these metrics, directly compare them, and assess the influence of subject group and data collection site on reliability estimates. It replicated previous activation study findings of reduced dlPFC activity during goal maintenance. It additionally found that the temporal association between a frontoparietal executive control network and a salience network was stronger in healthy controls than in schizophrenia patients and that the strength of this relationship predicted performance on the goal maintenance task. It also found that the task-modulation of the relationship between left- and right-lateralized executive control networks was stronger in healthy controls than in schizophrenia patients and that the strength of this task-modulation predicted goal maintenance task performance in healthy controls. Finally, reliability estimates found that ICA and tonic FNC had acceptable overall reliability and that they minimized site-related variance in reliability compared with dynamic FNC and general linear model. These results indicate that ICA and tonic FNC may provide better tools for group contrast fMRI studies examining schizophrenia, especially those that incorporate a multisite design.