Simulation of Blood Flow Through a Patient-Specific Carotid Bifurcation Reconstructed Using Deep Learning Based Segmentation of Ultrasound Images

One of the diseases of the cardiovascular system is the formation of carotid artery stenosis. The existence of atherosclerotic plaque within the vessel wall causes changes in blood flow and can have serious consequences to the individual's health condition. Therefore early and appropriate clinical diagnostics is very important. One of the first clinical examinations for this disease is the ultrasound (US) examination. Three-dimensional (3D) reconstruction and blood flow simulation could be used to overcome some of the drawbacks of the US examination and improve the overall diagnostics. An approach that combines the deep learning techniques and 3D reconstruction and meshing algorithms is applied within this study to first create the model of patient-specific carotid bifurcation and then to perform unsteady blood flow simulation, with realistic boundary conditions. This type of simulations can provide quantitative hemodynamic data to the clinicians during US examination and can further help to improve the diagnostics and ensure a treatment that is more adapted to the particular patient.