一种基于GPU的体积CT快速重建算法

A GPU-Based Fast Volume CT Reconstructive Algorithm Method

为了解决体积CT图像重建时间较长问题,提出了一种适合于医学临床应用的快速重建算法。首先,提出了一种基于图形处理器(GPU)的体积CT图像重建方式,利用GPU强大的并行和浮点运算能力进行计算效能的提升。其次,将体积CT图像重建中的几何运算与像素运算分离,减少了重复运算,进一步提高了计算效率。最后,基于医学应用背景,算法中实现了体积CT扫描和重建的并行化的思想。结果表明,利用上述的快速算法,在普通计算机硬件平台上即可实现重建时间减少70倍以上。

In order to solve the time-consuming problem in volume CT image reconstruction field,we brought forward a fast reconstructive algorithm method to fit the clinic medicine.First,using the great parallel and Floating-Point power of graphic processing unit（GPU）,we proposed a volume CT GPU-based reconstruction method.Second,in order to reduce the repetitive operation,we separated the geometry computation and pixel computation.As a result, the computation efficiency was further improved.At last,based on the background of medical engineering,the parallel of volume CT scan and reconstruction was implemented.The results demonstrated that the volume CT reconstruction time could reduce up to 70 times compared to the previous one in common PC by the fast reconstructive algorithm method.