A Hardware Architecture for Volume Visualisation Colour Imaging

This paper presents a hardware architecture for the real time generation of images from volume data. Colour imaging in this system is achieved though the use of a set of Look Up Tables (LUTs). The LUTs are capable of a variety of colour imaging techniques for visualising volume data. The use of colour imaging in generating images from medical volume data is discussed. Image synthesis from volume data in real time is an important technique in visualisation and graphics systems. INTRODUCTION The generation of two dimensional (2D) images from volumetric data sets containing either measured data or voxelized geometrical primitives is creating a new alternative for image synthesis techniques. The ability to synthesize 2D images in real time, that is at video presentation rate, will allow the interactive exploration of these data sets through several interfaces including Virtual Reality. Many aspects of real time volume imaging require improved technology, including processing power for image analysis and synthesis, and data sampling techniques for rapid image data acquisition in modalities, such as magnetic resonance imaging (MRI). Improvements in processing power can be achieved through the creation of special-purpose graphics imaging hardware [3]. Three dimensional sampled magnetic resonance (MR) data of the human body is used daily in the diagnosis of medical conditions. Real time MRI of the moving internal organs of the body will further assist medical treatment, in particular, heart conditions. Imaging technology of this nature will generate large amounts of data to be rendered for human inspection in real time. The technology described in this paper is intended for this type of application and also the rendering of voxelised geometric primitives. RELATED WORK Several systems for the generation of images from volume data in real-time have been designed and are capable of speeds of around one frame per second using supercomputers and custom architecutes. Comparable custom architectures to this system are currently being designed to handle large data sets at high frame rates. The Cube-3 architecture [5] is capable of processing a 5123 data set at 30 frames per second. Colour and opacity calculations in the Cube-3 system are precalculated and stored in Look Up Tables (LUT). Another system [4] capable of rendering images of 2563 data sets at 2.5 frames