Complex Networks-Based Exploration of Near-Wall Flow Disturbances in Computational Hemodynamic Models of Coronary Arteries

Cardiovascular diseases are a major cause of death , due to the rising incidence of obesity and diabetes in the Western world. Atherosclerosis is a vascular pathology where the accumulation of fat inside the arterial wall involves the formation of lipidic plaques, causing the narrowing or block of arteries. These plaques are usually located in the coronary arteries and atherosclerosis is in fact the main cause of coronary artery disease. The etiological factors of atherosclerosis have not been fully elucidated yet. However, it is well-known that hemodynamics plays an important role in the local onset and progression of vascular disease. This “hemodynamic hypothesis” suggests a co-localization of lesion prevalence at geometrically predisposed districts, such as arterial bifurcations, and disturbed flow regions, usually characterized by low and oscillatory wall shear stress (WSS). The aim of this thesis is to provide a better understanding of near-wall flow structures and hemodynamic risk by exploring the spatiotemporal evolution of WSS applying Complex Networks (CNs) theory. A dataset of ten swine-specific computational hemodynamic models of left circumflex (LCX) coronary artery was considered for this study. The method was applied to the time-histories, along the cardiac cycle, of the WSS vector magnitude (|WSS|). For each model, the Pearson correlation coefficient between each pair of nodal time-histories in the discretized fluid domain was computed, investigating the similarity between the dynamics of WSS in the coronary arteries. Later, each network was built by establishing a topological link between those pairs of time-histories correlated above a chosen threshold. CNs-based metrics were applied to the obtained networks to unveil their topological structure. The results from the CNs analysis were also compared to the well-established WSS-based descriptors of disturbed flow, to evaluate the ability of CNs in detecting flow disturbances at the arterial wall. The number of dynamically-correlated |WSS| time-histories connected to a node of the network is measured by the normalized degree centrality (DC). In the investigated LCX models, it emerges that the regions of the luminal surface characterized by low and oscillatory WSS exhibit very low values of DC: the WSS features in such regions are less strongly correlated with the hemodynamics in the rest of the arterial wall, thus resulting in a very small number of topologically connections. Our findings also showed a partial co-localization between disturbed flow surface areas and WSS patterns topologically isolated from the rest of the networks. By exploiting the network formalism, the here proposed approach proves able to describe the topology of the near-wall flow structures in coronary arteries and to capture its link with disturbed flow.