锥束CT图像重建算法
Image Reconstruction Algorithms for Cone-Beam CT
-
摘要: 随着螺旋多层面CT的出现,医用CT正在向着螺旋锥束CT转变。从螺旋锥束数据来重建图像有许多优点,但是这种成像方式在数学上比较复杂,技术实现也有相当的难度。本文介绍这一领域的重要成果,特别是近年文献中的基本思想。文中涉及各种主要重建算法,包括准确重建,近似重建和迭代重建算法。简言之,当数据是完整的、无噪音的,应选择准确重建方法;当数据是有噪音的和/或受物体运动干扰,近似重建会有较好的性能;而当数据是不可靠的和被截断时,迭代方法可以发挥作用。今后主要研究方向是对这些算法进行改进,比较它们的特性,并将其优点结合到一起。Abstract: With the development of Spiral multi-slice CT, medical CT is in transition towards spiral cone-beam CT. Image reconstruction from spiral cone-beam data is practically advantageous, but it is mathematically complicated and technically challenging. In the context of medical CT, the State of the at in the field is reviewed with an emphasis on fundamental ideas based on the latest literature. All the major types of reconstruction approaches are discussed, including exact algorithms, approximate algorithms, and iterative algorithms. Briefly speaking, the exact reconstruction approach should be the method of choice when data are complete and noise-free; the approximate reconstruction approach is often preferred when data are noisy and/or involving motion; and the iterative reconstruction approach may play a major role when data are corrupted and truncated. Major research efforts should be devoted to improvements, comparison and combinations of these algorithms.