锥束CT图像矩形包围盒快速重建

There is mass redundant computation in conventional cone-beam CT reconstruction. To solve the problem, fast reconstruction of rectangular bounding box is implemented. By prejudgment of object volume, the number of voxels to be reconstructed is greatly reduced. According to the analysis of the principles of Z-line data priority algorithm, it is feasible to conduct parallel reconstruction. GPU is then applied to accelerate back-projection computation. The experimental results show that the proposed method has higher computing speed compared with Z-line data priority algorithm and that the proposed method can be adopted in CT systems which employ analogous reconstruction algorithms.