光电工程, 2013, 40 (5): 97, 网络出版: 2013-05-24
基于GPU的超声弹性成像并行实现研究
Investigation of GPU-based Ultrasound Elastography
超声影像 弹性成像 应变成像 图像处理器 并行算法 ultrasound imaging elasticity imaging strain imaging GPU parallel algorithm
摘要
为了提高超声弹性成像计算速度 , 提出使用 GPU硬件加速基于互相关技术和相位零估计的弹性成像技术。先描述这两种弹性成像技术的实现细节及特点 , 然后分析这两种技术的计算密集操作部分的并行化计算可能性 , 最后通过 GPU程序开发工具 ArrayFire实现了基于 GPU的互相关和相位零估计的超声弹性成像技术。通过模拟和扫描仿真人体组织的弹性成像体模获得的压缩前后数据帧对基于 GPU的超声弹性成像方法进行测试与验证。实验结果表明, 基于GPU的方法可以大幅提高弹性图计算速度 , 在处理单帧弹性图条件下 , 与基于互相关方法比较 , 加速比达到 42×, 而基于相位零估计的方法在提高数据吞吐量的情况下加速比可达到 65×。
Abstract
In order to improve the calculation speed of ultrasound elastograms, two common ultrasound elastography methods based on Graphics Processing Unit (GPU) were investigated. After giving an introduction about the two common methods, the GPU-based approaches to rapidly estimate tissue strain was proposed. The two GPU-based approaches were tested in simulation signals and real signals by scanning a tissue-mimicking phantom respectively.Experimental results show that the GPU-based Normalized Cross Correlation (NCC) implementation is about 42 times faster than that based on CPU platform with the same elasticity image quality. At the same time, the GPU-based Phase Zero Estimation (PZE) approach is about 65 times faster than that corresponding implementation on CPU by increasing data throughput.
彭博, 谌勇, 刘东权. 基于GPU的超声弹性成像并行实现研究[J]. 光电工程, 2013, 40(5): 97. PENG Bo, CHEN Yong, LIU Dongquan. Investigation of GPU-based Ultrasound Elastography[J]. Opto-Electronic Engineering, 2013, 40(5): 97.