Fermi架构下超声成像组织运动可视化并行算法

在临床超声实时成像系统中组织运动情况是医生想要获取的重要诊断信息, 例如心脏运动. 基于线积分卷积的二维矢量场可视化技术可以同时展现运动矢量场的强度和方向. 但这一算法在处理时涉及大量的复杂计算, 尤其是流线追踪处理部分, 使其成为临床实时成像系统中的一大性能提升瓶颈. 为此研究并提出了一种基于新兴的高性能并行计算平台Fermi架构GPU(graphics processing unit图形处理单元)的并行运动可视化算法. 数据测试结果显示, 与基于CPU的实现相比, 采用Fermi架构的GPU处理不仅可

Parallel Processing Algorithm of Tissue Motion Visualization for Ultrasound Imaging on Fermi

On the clinical ultrasound real-time imaging system, tissue motion will be the important diagnostic information for doctors, such as cardiac motion analysis. In this paper, one motion visualization method based on line integral convolution (LIC) is studied. This method could show the vector field information about both direction and intensity. However because of the massive computation involved in this filter technique, especially in streamline tracking, it has been the bottleneck for the clinical real-time imaging system. In this paper, a new parallel algorithm of motion visualization based on Fermi GPU (graphics processing unit) is presented. Test results not only show the output of graphics processing unit (GPU) is definitely the same as the one of CPU, but also demonstrate the obvious speedup using GPU, that is, it can be 237 times faster than the CPU implementation with the long LIC filter (L=7) for the image size (260×260).