Institute for Data Analysis and Visualization

We present a method which allows us to visualize high-resolution medical volumetric datasets exceeding 512^3 data points at interactive frame-rates. Our approach is an out-of-core point-based rendering approach. In a pre-processing step we store the data points of the dataset sorted by their value on disk. At runtime we retrieve them when needed from disk to immediately stream them to the rendering hardware. The high resolution of the dataset and the density of the data points allows us to use a pure point-based rendering approach to achieve good image quality. The density of points with equal or similar values within the dataset can be considered as being high enough to display regions and contours using points only. Interactivity is not restricted to navigation through the dataset itself, changing the values of interest to be displayed can be done with real-time response. We are able to interactively display high-resolution datasets without any loss of detail. For datasets exceeding 512^3 values we are still able to display those datasets on standard off-the-shelf hardware. Using points only we reduce the amount of data to be transferred to the graphics hardware when compared to hardware-based volume-rendering or mesh-based approaches.