Impact of circular cross-section idealization in image-based computational hemodynamic models of coronary arteries

Coronary artery diseases are one of the most common causes of death in developed countries, in particular atherosclerotic stenosis caused by formation of fibro adipose plaques or calcifications, that lead to a narrowing in vessel lumen. This reduction leads to a decrease in perfusion of downstream regions and could subsequently lead to ischemic phenomena and myocardial infarction. Standard procedure for diagnosis of atherosclerotic lesions are invasive, but in recent years, image-based hemodynamic models obtained from computed tomography images combined with computational flow dynamics simulations is becoming a powerful tool, used to calculate descriptors useful to evaluate the fluid dynamic inside blood vessels and its effect on the atherogenesis. To properly reconstruct a three-dimensional model, a tomographic acquisition of images is necessary, while in the case of planar acquisitions, like coronary angiography, using epipolar geometry is possible to obtain a three-dimensional reconstruction with the assumption of circular cross section of vessel. The aim of this thesis is to understand the accuracy and the uncertainties associated to the assumption of circular cross-section, to do this, the same model was firstly reconstructed from computed tomography images provided by the Dipartimento di Scienze Mediche, Divisione di Cardiologia, Azienda Ospedaliera Universitaria Città della Salute e della Scienza, Turin, Italy, in particular left and right coronary arteries have been reconstructed. After that, using centrelines and radius information derived from the previous reconstructed models, the two models with circular cross section area have been obtained. To ensure grid independence and a lower computational cost, a mesh sensitivity study was performed. Prescribed boundary conditions were obtained from literature because in vivo measured data were not available for this patient. Unsteady-state simulations were performed and hemodynamics were described in terms of both surface and bulk flow descriptors, in particular, near-wall hemodynamics was quantified in terms of wall shear stress-based descriptors, while bulk flow was described in terms of helical flow metrics, evaluating the differences between CT-models and circular section area models. The goal of this study was to evaluate the approximation introduced by a reconstruction with circular cross-section area and its impact on the estimation of atheroprone areas. The results show that this approximation introduce a relevant underestimation of risk areas, this is surely dangerous because it can lead to neglect areas which are actually atheroprone, underestimating the atherogenic risk of the patient.