Biomechanical Soft Tissue Modeling - Techniques, Implementation and Application

The reaction of soft tissue to applied forces can be calculated with biomechanical simulation algorithms. Several modeling approaches exist. A scheme is suggested which allows the classification of arbitrary modeling approaches with respect to the degree of physical realism contained in the model (physical and descriptive models). Besides well known approaches like mass-spring, finite element, particle models and others the ChainMail algorithm is investigated. Where ChainMail in its original formulation lacked the capability to model inhomogeneous material, it is exceptionally stable and converges in one step to a valid configuration. In this thesis ChainMail is generalized to the Enhanced ChainMail algorithm which is capable to model inhomogeneous, volumetric objects and is fast enough for real time simulations. While now in principle being able to simulate and visualize an object in real time, a software architecture is required to team up simulation and visualization. As visualization and simulation have so far evolved independently, they work with different data structures. Multiplicity of data representations leads to the problems of data consistency and high memory consumption. A software architecture is developed which provides a universal data structure for several simulation and visualization approaches. The versatility of the developed architecture is demonstrated by two medical simulations. The first is the simulation of an intra-ocular surgery, which makes heavy use of Virtual Reality techniques. Designed as a training and educational tool the simulator EyeSi relies on descriptive real time ti me tissue simulation and visualization. The second deals with the simulation of decompressive craniotomy. The medical problem requires a physical model as the project's goal is to provide exact predictions on tissue behavior to support surgeons in surgery planning