Open Collections

Roes, Meighen M.

University of British Columbia

Recent decades of neuroimaging have witnessed a rapid increase in network-based research, reflecting the understanding that cognition arises from the interacting activity of multiple large-scale brain networks. However, research has increasingly prioritized studying how activity in resting state (RS) networks relate to cognition, rather than using task-based experimentation to observe the networks supporting cognitive processes. A central assumption of this approach is that task-based activity constitutes a ‘rebalancing’ of RS networks for task conditions. However, no research to date has adequately tested this assumption by evaluating RS and other network sets on their ability to explain task-related BOLD signal. In this dissertation, we introduce a novel method, Spatiotemporal fMRI-CPCA, and demonstrate its utility in characterizing the successes, and shortcomings, of different network models in accounting for the variance and spatiotemporal features of brain networks predictable from task timing. Through a series of quality assurance analyses, we validate the Spatiotemporal fMRI-CPCA method and illustrate the metrics by which network model performance can be evaluated. In order to establish a ‘ground truth’ against which network models could be evaluated, we measured and characterized the data-driven brain networks supporting verbal paired associates learning. Then, we employed Spatiotemporal fMRI-CPCA to examine how well two popular RS network models, and one network model using spatial templates derived from past observations of task-based networks, accounted for these data-driven networks. Overall, our results indicated advantages of the task-derived network model relative to both RS network models in capturing task-specific variance in BOLD signal, and in predicting key spatial and hemodynamic characteristics of the task-based networks. The advantages of the task-derived-templates model also extended to better accuracy characterizing differences in network activation between patients with schizophrenia and healthy controls. We discuss the implications of these findings for efforts to study the neural basis of cognition, and cognitive disorder, through observation of resting state networks.

Text

2022-01-12

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/80648

Psychology

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/