Generating geometrically deformable models by statistical shape modeling for computer aided dental restorations
Abstract
Automatic occlusal surface reconstruction for all kinds of tooth restorations is an important ongoing research topic. It is undisputed that an automation of the restoration process is only possible if the typical geometry of teeth is known by the system. One realizable approach is the restoration of the occlusal surface by adapting an appropriate tooth model. A first prototype that was developed at our institute is based on this idea. The main goal of this approach is the explicit use of dental knowledge in form of a small number of tooth models that have the ability to adapt to the patient's anatomy automatically. The range of occlusal surfaces is enormous and, furthermore, each tooth can vary in shape quite considerably between different patients. Any method of knowledge representation must be capable of representing a large class of flexible shapes but must also have sufficient specification to enable accurate reconstruction of the given tooth in real data. In this work, we describe the generation of such a tooth model based on a form of statistical shape analysis known as the Point Distribution Model (PDM). This includes an analysis of the shape variance and the definition of significant dental medical features. Then, this model can be used in our automatic CAD-system for dental restorations.
