| Since the late 20th century, network science has gained popularity in various fields, including the field of neuroimaging. Network science utilizes concepts of graph theory to model systems as complex networks. This approach is appealing to neuroscientists as it allows for a systems level investigation of the brain. While studies of human brain networks continue to mature, there is a need to extend this methodology to animal models, particularly nonhuman primates (NHPs). NHPs are considered a powerful translational tool that bridge basic science and clinical studies. They are also useful because the morphology of their cerebral cortex is similar to humans. In this work, a network processing pipeline was developed to build brain networks from NHP neuroimaging data. This model was designed to investigate the impact of long-term alcohol abuse in NHPs. Additionally, this work sought to understand what network analysis methods were the most informative for analyzing brain networks.;The first experiment describes the default mode network organization in NHPs. In addition, the effect of an acute alcohol challenge on NHP brain networks was assessed. It was determined that the default mode network of the NHP brain is similar to humans, but showed key differences, namely the inclusion of the superior temporal gyrus and visual cortex as highly connected regions of the network. Results of the acute alcohol challenge suggested that acute alcohol exposure disrupts the default mode network. The second experiment tracked longitudinal connectivity changes during a study of chronic ethanol self-administration in NHPs. Results from this study found that chronic alcohol abuse affects the default mode network in animals that drink heavily. Changes to the default mode network were linked to areas associated with spatial association, working memory and visuomotor processing. In addition to these results, analyses that modeled the consistency of network structure across a group were found to be more useful than averaging network properties. Moreover, despite weighted network analyses providing a more accurate representation of the network, their use proved limited as it did not change the interpretation of the results. |
