Many real-time volume visualization techniques use texture mapping, which has several disadvantages. For example, 3D texture depends mainly on regularly sampled data on texture coordinates such as 3D rectilinear grid. Other techniques depend on regular proxy geometry to map texture. While texture mapping exploits hardware acceleration in graphics processing units (GPU), they introduce approximation artifacts in ultrasound imaging applications with polar sampling. Alternatively, ray casting visualization does not depend on proxy geometry to map data on. This alleviates approximation artifacts introduced by texture mapping. The problem with ray casting is its prohibitively high computational complexity especially in real-time applications such as 4D ultrasound. Recent GPUs support programmable pipelines and multiprocessing streaming units, which make it possible to implement ray casting in real-time using GPUs [2].