Geometrical Reconstruction And Simulation Of Hemodynamics For Patient-specific Intracranial Aneurysm With Lattice Boltzmann Method

Intracranial aneurysm is one of the most common and dangerous diseases that cause serious harm to human health, and hemodynamic characteristics, as a main factor in the formation, growth, rupture and cure of intracranial aneurysms, have drawn wide attention. Intravascular interventional treatments such as intracranial stent-assisted angioplasty and coil embolization have achieved good effect in the clinical aspect, however, the design and implementation of medical equipment still need to be supported by theoretical and exper-imental evidence. Traditional clinical trial is in the midst of difficulties of high risk, long period and high cost, therefore numerical experiments become the main method of research on hemodynamic. The blood fluid properties, the complicated geometry of blood vessels, different scales between the vascular and the stent, as well as large-scale computation limit the traditional numerical methods in studying the hemodynamics of intracranial aneurysms.During the past two decades, the lattice Boltzmann method (LBM), as a kinetic based numerical approach, has gained a great success in the study of the complex flows and also can be viewed as a general solver to some partial differential equations (PDEs), such as convection diffusion equation, reaction diffusion equation, Possion equation et.al.Recently, the LBM has been used to simulate hemodynamic of intracranial aneurys-m by many researchers. But most of their work are based on ideal and simple geometry model, which are assembled with cylinder and sphere. The initial and boundary conditions for the patient-specific model still need to be further discussed. Based on this situations, the aim of this paper is to study the whole process of the hemodynamic simulation of the patient-specific intracranial aneurysm with lattice Boltzmann method. Based on the comput-er tomography (CT) medical imaging data, through developing advanced lattice Boltzmann model and high-performance computing technology, we solve image segmentation,3d ge-ometry reconstruction, the LBM computing grid and hemodynamic simulation.Geometry reconstruction of the patient-specific intracranial aneurysm and LBM com-puting grid:(1) We present a modified LB model as a general solver to some PDE-based image processing models, and use it to image smoothing, edge detection and segmentation of the patient’s CT medical imaging data. The experiment results show that our LB model can ac-curately segment the three-dimensional geometric model from the background. The method is not only fit for medical image processing but also for more complex image processing task, such as filtering and contour detection of the natural images. (2) In order to meet the needs of the LBM computing grid and the study, we put for-ward some approaches to improve the existing geometric model, that make the geometric reconstruction result closer to reliable. We propose a vascular centerline method to recon-struct the tumor vessel and stent. On the one hand, the tumor vessel becomes more smooth, which is more close to the actual vessel. On the other hand, we can increase or decrease the branch vessel near the aneurysm according to demand of the experiment. Comparing with traditional CFD software computing grid, the proposed method is more reasonable and flexible for the design of the stent.Hemodynamic simulation of intracranial aneurysm with complex boundary:(1) Based on the proposed geometry reconstruction approach, we extend complex boundary lattice Boltzmann model to the Navier-Stokes equation, which can describe the hemodynamic of clinical patient-specific intracranial aneurysm. We analyze the boundary conditions of the inlet and outlet, and found that velocity inlet and pressure outlet are fitter for various patient-specific geometrical model. Based on this boundary condition, we discuss differ-ence between the pulsation velocity and constant velocity. The blood flow form at the time of the pulse peak is similarity to that take the average velocity of the mean value and max value as the inlet boundary condition.(2) We choose several clinical cases of intracranial carotid aneurysm (ICA), and study the differences of hemodynamic parameters before and after the stent implantation. The ge-ometric models with and without branch vessel are also studied. According to the numerical results, the magnitude of velocity in the aneurysm obviously decrease after the stent im-plantation, and the reduction is in direct proportion to the mesh density of stent. When the branch vessel is not close to the aneurysm or the diameter ratio of main and branch vessel exceed4, the velocity and flow field change relatively little.In conclusion, by using the lattice Boltzmann method as the main numerical method, we establish a whole process of hemodynamic simulation for patient-specific intracranial aneurysm and use it for a mass of clinical patients. Different geometric conditions are considered, which establish the basis for further research on intracranial aneurysm and provide convenience for exploring the treatment. Furthermore, the GPU based LB is further developed to improve the computational efficiency of LBM in our study, and the results demonstrate that our parallel GPU algorithm achieves about two order of magnitude faster than that of the CPU based algorithm. We attempt to place a simple helical stent into the parent vessel, the numerical experiments are consistent with clinical practice. This hemo-dynamic simulation approach makes the design of patient-specific stent and its position in the vessel become more feasible and reasonable, which promotes the development of clinical treatment.