Scalable volume rendering

Background

Visualization of medical data across a large range of computing systems requires a scalable rendering framework. Computing systems ranging from Mobile Platforms used for informal discussion of medical data through to high-end systems providing detailed Visualization can be utilized with a single scalable rendering system. Utilizing state-of-the-art research in multi-resolution volume rendering, medical data can be compressed for a wide range of systems and network bandwidth requirements. However, this only solves the data transmission part of the problem. A major missing component is the capability of the system to adapt to different platforms with a wide range of computing power. Often visualization systems are designed for a single device, and when technology changes, the system needs to be rebuilt and redesigned. Hence, we propose to develop and investigate new algorithms for this purpose.

Objectives of the project

Our approach is to develop a scalable rendering system with an integrated model of computational capability for graphics which enables finely tuned visualization using the current computing system. We will develop a model of computational capability that is future proof by representing a wide variation of architectures from low-end GPUs found in mobile devices, right through to high-end GPUs and upcoming parallel architectures such as Intel’s Knights Ferry and NVIDIA’s Fermi. The advantages of such a system are many. In contrast to current fixed system solutions, a scalable model can easily be updated as new architectures are introduced. In addition, as high-end architectural concepts migrate to mobile platforms the rendering system is already equipped to handle the new device.

Project Members