Computational Simulation of Platelet Transport, Activation, and Deposition

Platelet-mediated thrombosis is a significant source of morbidity and mortality in cardiovascular device patients. Although Virchow elucidated the mechanisms governing thrombus formation over 100 years ago, the underlying processes have proven difficult to describe mathematically. A reliable, predictive thrombosis model would be a valuable aid to designers of artificial organs. A comprehensive model of platelet-mediated thrombogenesis should simulate red-cell-enhanced platelet transport, platelet activation, kinetics and mechanics of platelet deposition and aggregation, flow disturbances due to thrombus growth, thrombus disruption by fluid forces, and interactions between platelets and the coagulation cascades. Most models focus on these components individually; a unified approach is lacking.The contribution of this thesis is a computational model of platelet thrombosis which incorporates many, though not all, of these essential components. This two-dimensional continuum model, based on prior work, is comprised of seven coupled species conservation equations, which model shear-enhanced platelet transport, platelet-platelet and platelet-surface adhesion, agonist-induced platelet activation, platelet-phospholipid-dependent thrombin generation, and heparin-catalyzed thrombin inhibition. The model is first validated for Poiseuille flow of whole human blood over collagen. Very good agreement between predicted and experimentally measured platelet deposition is obtained for wall shear rates ranging from 100 to 1000/s. At 1500/s, however, the model fails to predict the shape of the experimental curve. This may be due to the higher shear rate, or to unmodeled effects of the chelating agent used as the anticoagulant in this study.Next, two-dimensional flow over collagen is considered. For a tubular expansion with no agonists, good agreement with experiments can be obtained in the recirculation zone, but deposition in the fully-developed downstream region is greatly under-predicted. Surprisingly, deviation from experiments is worse at 0% than at 20% hematocrit. Similarly, for an axisymmetric stenosis with agonists present, the model does most poorly at predicting deposition in the downstream regions. These discrepancies are attributed to the approximation of blood as a single continuum, rather than as a suspension. In a preliminary step toward correcting these deficiencies, an attempt is made to use existing two-phase flow models to predict platelet and red blood cell concentrations in fully-developed tube flow