Study On The Visualization Of Ultrasound 3d Reconstruction And Improvement Computer Aided Diagnosis System (CADS)

In 1950s, ultrasound was the introduction to medical uses. Compared with other medical imaging techniques (CT, MRI), ultrasound imaging has its unique advantages: real-time, interactive, non-electromagnetic radiation, non-invasive, economical and convenient, easy to repeat the examination. At present, the clinical applications of ultrasound are mainly two-dimensional. With the rapid development of computer technology, there are increasingly stronger calls for three-dimensional ultrasound. Compared with the two-dimensional ultrasound, three-dimensional ultrasound has the following advantages: first, the direct three-dimensional structure of organs avoid over-reliance on experience of doctors; second, it can give any direction of the two-dimensional surface images, some of which can not be obtained by two-dimensional ultrasound image, and these directions often play a key role in the diagnosis; third, measure an organ more accurately; four, can be kept for further study; five, intuitive and easy to understand.Ultrasonic three-dimensional reconstruction can be divided into three stages: image acquisition, reconstruction, and display. Ultrasound reconstruction reconstructs a series of two-dimensional ultrasound image into three-dimensional image based on scanning probe position and orientation. Currently, large overseas companies have developed a number of ultrasound scanning and reconstruction equipment. But they are expensive, not real-time, and limited in scanning range, so they are not popular in China. Thus, we have developed algorithm that is based on two-dimensional ultrasound equipment and general computer, real-time, and free hand. It's inexpensive, easy to spread, and also broke through the scanning range limit.At the same time, there are several methods for three-dimensional visualization of medical data, such as the Marching Cube for surface display, Volume Rendering for direct visualization. However, with the poor quality ultrasound image is almost impossible to extract any surface, it can only be displayed with volume rendering. However, the most popular method of volume rendering Ray Casting needs large amount of calculating, so difficult to achieve real-time display. Despite previous study, there are still many points can be developed, especially how to display real-time on general-purpose computer platform. To solve this problem, this paper has developed Ray Casting method based on CUDA (Compute Unified Device Architecture).This article discusses the ultrasound acquisition, reconstruction and real time visualization of the reconstruction results in three-dimensional. It is divided into the following parts: the ultrasound image acquisition and reconstruction, three-dimensional visualization based on GPU-accelerated, visualization optimization for three-dimensional ultrasound data, and in the experimental section, combination of ultrasound reconstruction and three-dimensional visualization is discussed.In ultrasound acquisition and reconstruction, first we discussed how to use the image acquisition card to obtain two-dimensional ultrasound images, using magnetic locator to obtain the location of each frame ultrasound images. Then we discussed the reconstruction of pixel-based methods of implementation, as well as how to deal with the reconstruction holes. And then we explored several methods to accelerate the reconstruction, such as using the characteristics of ultrasound images, multi-threaded and so on. Finally, we explore how to display the reconstruction process real-time, so physicians can make some adjustments during reconstruction process.In visualization of three-dimensional data based on GPU, we introduce Ray-Casting based on CUDA. First we introduced how to read the three-dimensional data into system memory and deployed to GPU. Then we introduced the calculating deployment, that is, how to arrange so many threads to each of GPU core reasonably. And then focus on the kernel of Ray Casting algorithm, and come up with several factors that may affect the display quality and speed, such as the direction of the sampling in depth, early-terminate transparency and so on. Finally, discusses how to display the computation result.In optimization for ultrasonic three-dimensional display, we study the difficulties of optimization, and argue that only if users participate in the process of visualization, that is, the mapping process from the intensity to RGBA, three-dimensional display can meet the needs of users. Following this point, we first design a friendly GUI to determine which features should be contained. Then, we designed the software architecture, and define several projects and dozens of classes. Finally we analyze several typical processes.In the experimental section, we first show several of ultrasound reconstruction results, and then analyze how the various parameters of the three-dimensional display affect the image quality and speed. Finally combined ultrasound reconstruction and three-dimensional display, display the reconstruction process in three-dimensional.