锥束CT的改进FDK算法及加速实现  

Improved FDK Algorithm and Accelerated Implementation for Cone Beam CT

作　　者：徐月晋 陈梓嘉[2] 骆毅斌 胡洁 

机构地区：[1]广州华端科技有限公司,广州510530 [2]南方医科大学生物医学工程学院医疗仪器研究所,广州510515

出　　处：《核电子学与探测技术》2015年第11期1124-1127,共4页Nuclear Electronics & Detection Technology

基　　金：国家科技支撑计划课题(2015BAI01B10)资助

摘　　要：锥束CT成像中,FDK重建算法得到广泛应用。提出一种可以扩大锥束CT旋转轴方向重建体积的FDK方法。在反投影步骤中,对重建物体的每个体素点被反投影的总次数进行计算,之后进行权重计算,最终得到三维重建图像。同时,采用CUDA技术,利用GPU对改进的FDK算法进行并行计算加速实现,提高重建效率。实验结果表明,改进的FDK算法可以增大重建体积,重建更多物体信息,采用CUDA技术后,速度大幅度提升。FDK algorithm has been applied widely in cone beam CT imaging. In this paper,we have proposed an improved FDK method which can enlarge axial reconstruction volume. In backprojection process,the backprojection times of each voxel need to be counted then take part in weighting calculation. At last,reconstructed image is obtained. Meanwhile,CUDA technique using GPU is used to accelerate the implementation for proposed FDK algorithm. Experimental results have demonstrated that,the improved FDK can enlarge reconstruction volume and reconstruct more object information.

关 键 词：锥束 断层成像 FDK算法 图形处理器 加速实现 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]