基于GPU的光栅相衬成像的相位恢复和CT重建
陈晓庆1，王宇杰2，孙建奇3
基金项目：教育部博士点基金新教师类项目基金（20100073120004）；国家973项目基金（2010CB834300） 作者简介：陈晓庆（1989-），女，硕士研究生，光栅相衬成像系统的相位恢复和CT重建算法研究 通信联系人：孙建奇（1981-）， 男， 助理研究员，博士， 同步辐射医学影像.

（1. 上海交通大学生物医学工程学院，上海 200240； 5 2. 上海交通大学物理与天文系，上海 200240； 3. 上海交通大学生物医学工程学院， 上海 200240）
摘要：在医学图像领域中，区分吸收系数相近的物质的要求越来越高，同时也越来越普遍，如血管和组织，血管和肿瘤等。相比传统基于吸收信息进行重建的方法，光栅相衬成像的方10 法利用不同物质相位信息变化更为明显的特点对样品进行重建获得了更好的图像质量。但随之而来的是，光栅相衬成像产生了更多的实验数据需要处理。通过对GPU的CUDA C并行编程，将相衬成像数据在GPU内核函数中进行相位恢复和滤波反投影重建取得了既快又好的实验结果。光栅相衬成像实验结果显示出了很好的空间分辨率和对比度分辨率，尤其是针对弱吸收的软组织，跟传统的吸收成像相比，具有更强烈的对比度和更清晰的组织结构，在以后15 的应用过程中能够帮助医生更准确的找准病灶并分析。其次，相比于标准C程序实现CT重建，基于GPU的CUDA C的光栅相衬成像CT重建程序获得了比较明显的加速比，尤其是实验数据增多的情况下，加速比的增长趋势也变得更为明显和突出。
关键词：生物医学工程；光栅相衬成像；GPU并行加速；滤波反投影 20
中图分类号：R318.6
GPU-Based Phase Retrieval and CT Reconstruction for Differential X-Ray Phase Contrast Imaging
CHEN Xiaoqing1, WANG Yujie2, SUN Jiaoqi3 25
(1. School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai 200240; 2. School of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai 200240; 3. School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, 200240)
Abstract: In this paper, compared with the conventional absorption imaging, the differential X-ray phase contrast imaging shows a higher spatial resolution and contrast resolution on reconstruction 30 images especially when dealing with the samples with low attenuation coefficient such as the vessels, soft tissues and so on. The grating based phase contrast system has been widely used with its advantages on the wide range of sample selections and independence with the crystal compared with the propagation-based phase contrast imaging and the diffraction enhancement imaging. However, large data acquired from the phase-stepping method in every projection angle consumes 35 the time of reconstruction. Graphic Processing Unit (GPU) has a wide advantage on parallel computing with multiprocessors and thousands of threads. This paper introduced the CUDA C programming model based on GPU to accelerate the phase retrieval and FBP algorithm on grating phase contrast system. According to the different size of data, the CUDA C program shows a different speed-up over the standard C program on the same Visual Studio 2010 platform. 40 Meanwhile, the speed-up ratio shows an increasing tendency when the size of data increases
Key words: Biomedical Engineering; Grating-based phase-contrast imaging ;parallel computing based on GPU, Filtered Back Projection