As computer power and graphics capabilities increase, there is growing interest in surgical simulation as a technique to enhance surgeons' training. A computer simulation system could potentially save time and money, reduce the need for cadavers and laboratory animals, and provide objective feedback. However, for surgical simulation to be a useful training tool it must be realistic with respect to tissue deformation, tool interactions, visual rendering, and real-time response.; This dissertation presents several algorithms to deform objects realistically, and interact with these objects in a simulation environment. Soft tissues are represented as mass-spring meshes, allowing for dynamic simulation, as well as a new quasi-static algorithm which improves the simulation speed, provides a guaranteed frame rate, and scales up to large meshes via an automatic computation cutout. Precise collision detection and response are necessary for creating realistic interactions between various types of deformable and rigid objects, and we discuss a simple but efficient hierarchical collision detection scheme. Our work also introduces a novel method of suture (or rope) motion, which looks natural and obeys the physical constraint's caused by both self-collisions in the suture, and collisions between the suture and other objects. We describe what we believe to be the first real-time knot tying simulation, with automatic identification of the knot types.; Finally, we present a specialized microsurgery application which combines and extends the above algorithms. This simulation resolves the interactions between forceps, a suture, and deformable vessels. Using a tracking device to control the forceps, a user can manipulate the suture through two vessel ends, pull them together, and tie them off using the appropriate knots.
展开▼
