FAST PHYSICS-BASED SIMULATION OF VASCULAR SURGERY

Abstract

Minimally invasive vascular surgery has been shown to be effective for the treatment of vascular or artery diseases. However, extensive training and careful planning are needed for a successful operation due to its inherent problems including non-co-located hand-eye coordination, restricted vision of anatomical environment, and the absence of tactile feedback. Virtual reality medical simulation can potentially provide a safe and robust solution to the training of clinicians.

In order for the training to be effective, the simulation should be physics-based with real-time computer graphics rendering so as to be realistic. This dissertation focuses on the study of fast physics-based simulation of vascular surgery on three important aspects: (1) realistic and interactive simulation of blood vessel deformation using an improved lumped element method, (2) fast hemodynamic simulation in patient-specific model using meshfree methods, and (3) simulation of surgical smoke during the surgery.

Human blood vessel can be regarded as a soft-bodied object. A simulation of blood vessel deformation should allow users to interactively manipulate a blood vessel model based on the biomechanical properties of the tissue. The simulation should resemble the soft tissue division process where the blood vessel can be torn into several parts. Therefore, an improved lumped element method is proposed to simulate the blood vessel deformation. The blood vessel model is segmented and reconstructed from clinical CT-images. The method achieves good simulation realism at high computational speed. Real-time simulation and interactive control are demonstrated in an application on patients with abdominal aorta aneurysm by using GPU for general computing.

Study of hemodynamics is normally based on existing cases of patients with specific anomalies. Evaluations and investigations of new cases are limited in practice due to the need for accurate and timely prediction of blood flow. Hemodynamic simulation of blood flow based on the reconstructed patient-specific model is challenging. In this dissertation, two meshfree methods have been used to simulate the blood flow and drug delivery in patient?s vascular system. The first approach is based on an improved Smoothed Particle Hydrodynamics (SPH) method which is designed to render a 3D graphical simulation. The second approach is based on an improved Finite Particle Method, and the method was implemented in two cases of blood flow simulations with respect to hand and heart circulation with blockages. The meshfree method avoids the computationally intensive meshing process in conventional hemodynamic simulation. Tensile instability of the SPH formulation is reduced with a smoother approximation of second derivatives in the system. Mass and momentum conservations are achieved locally and globally in the continuum equation. The simulation results have shown to be efficient and realistic for the prediction of blood flow in patient-specific vascular model. Surgical smoke is generated during the ablation of blood vessel in vascular surgery. It is inevitable that clinicians and patients in operation theatre exposed to the pollution of surgical smoke plume created due to the thermal destruction of tissue. In this dissertation, an improved vortex particle method is used to simulate the dynamics of surgical smoke. The method couples the advantages of both grid-based and particle-based methods. The coarse mesh in the grid-based method is used to predict the main flow trajectory of the smoke fluid, and the vorticity force obtained in the particle-based method is employed to restore the lost details due to numerical dissipations. The realistic and robust simulation can be used for medical training and planning.