Research In Soft Tissue Modeling And Collision Detection Technology Of Virtual Surgery Simulation

Virtual surgery technology is an important application of virtual reality technology in the medical field, it is the cross-over study in the field of computer science, biology, medicine,mechanics and other disciplines. Because of its prominent role in the training of surgery,preoperative planning, it gets more and more attention in the development of modern medicine. Virtual surgery training system is an important application branch of virtual surgical technique, which uses a computer to create a virtual environment of real surgery training scenarios, the medical trainer operates the surgical objects with surgical instruments controlled by the interactive device in the scene,virtual environments provide a realistic feedback to the trainer in the visual and tactile sensory to make the medical trainer feel like training in reality. Such systems can reduce training costs and improve training effectiveness in training of medical personnel. In such a system, the critical and difficult technology are the physical modeling of soft tissue and collision detection,these two technologies are the key factors affecting the real-time and realism of virtual surgical simulation training systems , this paper did depth research in both of them.Firstly, we do some research in the physical modeling technology of soft tissue. Mass -spring model and the finite element model are the popular model used in traditional virtual surgery training system. Mass - spring model does not require continuous physical parameterization, easy to implement; when dealing with cutting and sewing operation, the removal and increase of nodes are easy; but its deformation accuracy and stability are poor.Finite element model has the advantage of high precision; but there is a huge amount of computation and have difficulty to handle the topology structure when it changes. For the above-mentioned problems, this paper will apply meshless method based on shape matching to build the physical model of the soft tissue, this method does not require meshing, there is no connection between the particle, it is simple and easy to implement, unconditionally stable,and can achieve satisfactory results on the deformation speed and accuracy aspects.Secondly, this paper reseach in how to improve the efficiency of collision detection too.According to the characteristics of different bounding boxes and objects in the scene, this paper proposes to use Sphere-AABB complex bounding box in collision detection which overcoms the shortcomings that the accuracy and timeliness cannot be solved well when using a single defect. At the same time this paper improves the collision detection algorithm for base primitive (triangle patch) based on projection proposed by Yu Haiyan to enhance the efficiency and accuracy of collision detection.Finally, to build a simple liver virtual surgery training system by use of the above physical modeling and collision detection technology to do simulation experiments. In the scenario the mean value of fps is over 50 and visual feedback is smooth,which meets the real-time requirements. The deformation of pushing and pulling of the liver are closer to the real deformation effects, there is no collapse or super-elastic phenomenon, at the same time surgical instruments are not far from or cross the soft tissue, which meets the requirements of verisimilitude. So the technologis of physical modeling method based on shape matching meshless theory and collision detection method based on complex hierarchical bounding box tree are effective and practical, they can be applied to the actual virtual surgery training system.