锥束CT的改进FDK算法及加速实现

Improved FDK Algorithm and Accelerated Implementation for Cone Beam CT

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.