Our system designs automatically collision-free restorations which also meet the requirements imposed on the load distribution over teeth. The shape of the occlusal surface is ``reasonable'', i.e., main morphological features of teeth are preserved and the rules of esthetic dentistry are respected. The dentist can examine images of a 3-D surface of a restoration, and CDM with the minimal distance to the opponent teeth and the load on the restoration. In some cases, especially for patients with occlusion disorders further adjustments are necessary. Also, if a dentist dislikes a surface proposed by the system for, (e.g., esthetics), the designed restoration can be interactively changed.
The interaction method is inspired by the traditional wax up technique for design of crowns and dentures (Payne/Lundeen, Thomas) [Oke89]. This method is widely used by dental technicians. The modeling takes place in a mechanical articulator. The crown or denture is designed by adding wax layer by layer on a gypsum model which is gotten via imprint. The warm wax is transparent and easy to modify. As it cools (e.g., accelerated by blowing), it looses its transparency and becomes hard. During modeling the design is checked with the articulator. For visualization of the contact points a colored powder is used. The powder is sprinkled on one side of the jaw in the area of interest. If the teeth get into contact during manual sliding, the powder is transfered to the opposite tooth.
Virtual reality provides many advantages over manual design. Teeth which are not involved in the design can be hidden. As the jaw either moves along recorded trajectories or is controlled by the user, contacts and collisions are directly visible as distance maps mapped on the teeth. Also, watching the penetration from the back of a tooth or through a semi-transparent opposite tooth is possible. Only virtual reality allows us to go into rigid bodies and discover the whole nature of the articulation process.
In the modeling part of our system we simulate the behavior of wax. The dentist chooses the amount of virtual material which can be used during one step of the editing. This limits the surface area and its relative changes. The editing process starts by touching the surface which triggers a clock. A control panel, represented as a cylinder which change its color from red (hot) to blue (cold), indicates the time (temperature). Depending on this simulated temperature, changes corresponding to mouse or stylus motion are scaled down. The process of cooling down can be speeded up (blowing) by user interaction. The previous shape is visible during editing, because the new surface layer is translucent as long as it is modifiable (see Fig. 3). An undo operation is available. After completion of the shape design, the occlusal surface is checked against collisions using the distance maps technique.
For our application, we propose specialized software (manipulators),
based on direct
manipulation of objects [CSH
92].
Conventionally used scrollbars, keyboard or mouse interfaces are not suitable for
editing in three dimensional space. The mouse becomes more effective
for editing when
attached to a 3-D widget with knowledge about the modified object.
The software manipulators have suitable hardware counterpart in the form
of control devices (spaceball, stylus, abstract mechanical model).
An abstract mechanical model of the jaw helps
the dentist to determine orientation in 3-D space. A similar approach
has been used to do
neurosurgical planning [GHP
95].
A doll's head and stylus are input devices to
control editing and visualization. This allows reduction of the cognitive
distance between user and model. Our metaphors are based on
traditional approaches using
mechanical articulators and stylus, with a dialog technique based upon
direct manipulation of objects.
Disadvantages of our prototype compared to a real world
approach includes the lack of force feedback and that the user sometimes
loose the feeling for the size of his changes.
