Design and Development of a VR System for an Exploration of Medical Data Using Haptic Rendering and High Quality Visualization
Roman Vlasov, Leibniz Universität Hannover , 2016
Haptic exploration adds an additional dimension to working with 3D data: a sense of touch. This is especially useful in areas such as medical simulation, training and pre-surgical planning, as well as in museum display, sculpting, CAD, military applications, assistive technology for blind and visually impaired people, entertainment and others.
Each haptic rendering frame consists of three stages: collision detection, collision response and force feedback generation. In order to feel the 3D data smoothly, an update rate of at least 1~kHz is required. There exist different surface- and voxel-based haptic rendering methods. Unaddressed practical problems for almost all of them are that no guarantees for collision detection could be given and/or that a special topological structure of the objects is required. Here we present a novel and robust approach based on employing the ray casting technique to collision detection and path finding for collision response. The approach is very fast (150~kHz) and does not have the aforementioned drawbacks while guaranteeing nearly constant time complexity, independent of data resolution. This is especially important for delicate procedures, e.g. pre-operation planning. The collision response uses an implicit surface representation, which can be used with dynamically changing objects, as no precalculation is needed. Further on, we present our flexible deformation framework allowing us to use our haptic rendering approach together with deformation models. We present our graphics approach which we use to keep the graphics representation of segments up-to-date during the deformation simulation. The challenge here is to reflect deformations of objects interactively. Further on, we propose two local deformation simulation approaches based on the method of potential fields. The first approach uses "regular'' potential fields. The second approach uses our novel cuboid fields. Further on, we demonstrate that cuboid fields are better suited to haptic rendering of volumetric data. Additionally, we introduce the prototype of the global deformation approach. The resulting haptic rendering approach combined with our proposed approaches for deformation simulation within our deformation framework does not require any pre-calculated structure and works "on the fly"'.
Our deformation framework and all our haptic rendering and deformation simulation approaches were fully developed by us from scratch. Their design and development was the main aim of this work. This project was supported by a grant provided by Siemens/DAAD Postgraduate Programme.