Planning and rehearsal of surgical interventions in the volume model.

Visualization of 3D medical data is routinely used in a wide range of applications. However, for the planning and rehearsal of surgical interventions more sophisticated techniques for interaction have to be developed. The realistic specification and visualization of free form cuts is needed to allow the 'look and feel' close to a real dissection. The problem here is, since these cuts are not represented by intensity changes, that the gray-level-gradient-method can not be used for the estimation of surface normals. In addition, the interactive repositioning of dissected fragments has to be simulated. We have developed an extended ray-casting algorithm for visualization of object motion in the volume model. We implemented new methods for the representation, modeling and high quality rendering (subvoxel resolution) of arbitrarily shaped cut regions within the volume model. The representation is done using a dynamic data structure. This way, all operations can easily be reversed and the original object information is preserved. The modeling of cut surfaces is done in an independent data volume where the partial-volume-effect, which is the prerequisite for the gray-level-gradient method, is calculated as it would be generated by an imaging system. This way, the localization of cut surfaces at subvoxel resolution and an accurate estimation of the surface normals is achieved. The key point here is to detect if a cut surface really truncates an object or if the object has not been affected by a cutting operation. We will present an new method, called adaptive sampling which allows to determine the situation by the generation of additional sample points (when necessary) during the ray casting process. The described techniques provides the basis for simulation of surgical interventions in the voxel-model which could not be achieved with any surface-based method. We present a system for simulation and rehearsal of otosurgical approaches, where we implemented a drill-like tool with which the student lays off the route to the operating area. The key point is to not injure structures of risk such as the facial nerve. For applications like the simulation of craniofacial surgery we developed a gradual cutting tool ("virtual scalpel").