Four-dimensional (4D) ultrasound imaging extends the real-time capability of ultrasound to visualize a real-time volume that can be manipulated by the sonographer. Among the different visualization methods, surface rendering is a common mode for displaying volumetric datasets such as in obstetrical applications. A challenge in this mode is that surface shading is required to visualize the surface and enhances the surface contrast and this has very demanding computational requirements for 3D surfaces. Due to the low quality of the 2D ultrasound images as a result of the presence of speckle noise, we develop a new real-time methodology for processing of the 2D ultrasound images before being used in the rendering pipeline. Then, the thesis addresses the development of an optimized high-performance rendering pipeline based on four stages for preprocessing, volume rendering, surface shading, and post-processing. The new approach is implemented to render real ultrasound volumes on a 4D commercial ultrasound imaging system to verify and validate its performance and illustrate its practicality. The results demonstrate diagnostic quality of rendered volumes at a computational time cost that is suitable for 4D real-time processing requirements, given its low cost of required hardware; the new pipeline has potential for making 4D imaging systems more affordable while maintaining diagnostic quality and performance. |
