Neurosurgical Simulation System with Octree-based Implicit Volume Modeling

摘要

This paper proposes a neurosurgical simulation system with a implicit volume model represented by parametric continuous functions and a haptic sensation. Constructing an implicit smooth surface that is used for feedback force calculation and polygon surfaces from the function representation of voxel model, only a specific organ can be visualize by both volume rendering and surface rendering. In this study, a 3D head model is constructed by CT image slices. The head model is a correction of voxels, and each voxel has color and density values at discrete location. Volume rendering visualizes an inside of a human head in detail. Voxel expression, however, has the following problems; unevenness of a feed back force, roughness in enlarged image, and difficulty of handling a specific organ or tissue. To overcome above problems, we introduce approximating method based on the Octree and tri-cubic parametric volume function. In this method, the voxel data is approximated locally by means of the parametric continuous functions. In calculating a feedback force, implicit surfaces are generated by extracting points on surface at level zero. The parametric surface approximating extracted points is locally created. Polygon surfaces are created by applying the Marching cubes to our system. In rendering the polygon surface, zero crossing points are an extracted from the continuous function efficiently, and constructed surfaces are visualized. When the shape of extracted organ is visualized, the boundary surface of the organ is rendered with surface rendering. On the other hand, when the inside of an organ is visualized, the internal structure is rendered by volume rendering. In addition to realize smooth tactile sensation and smooth surface rendering, the see-through and fly-through functions are equipped to diagnose the patient head and to plan the operating path. The experimental results of the polygon surfaces, the parametric surface, and the simulation demonstrate the usefulness of our system.