一种投影数据恢复的扇束CT局部重建算法  

A local reconstruction algorithm of fan beam CT for projection data restoration

作　　者：梁亚星 张权[1,2,3] 张萌 陈燕[2,3] 桂志国[1,2,3] LIANG Yaxing;ZHANG Quan;ZHANG Meng;CHEN Yan;GUI Zhiguo(National Key Laboratory for Electronic Measurement Technology,North University of China,Taiyuan 030051,China;Key Laboratory for Biomedical Imaging and Big Data of Shanxi Province,North University of China,Taiyuan 030051,China;College of Information and Communication Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学电子测试技术国家重点实验室,山西太原030051 [2]中北大学生物医学成像与影像大数据山西省重点实验室,山西太原030051 [3]中北大学信息与通信工程学院,山西太原030051

出　　处：《中国测试》2021年第5期90-96,共7页China Measurement & Test

基　　金：国家自然科学基金(61671413,61801438);山西省自然科学基金项目(2015011046,201901D111153);山西省青年科学基金(201801D221196);电子测试技术重点实验室开放基金项目(ZDSYSJ2015006);中北大学青年学术带头人项目(QX201801)。

摘　　要：针对局部扫描过程中投影数据沿探测器方向截断的特点,提出一种基于投影数据恢复的扇束CT局部重建算法。算法采用线过程模型检测截断投影数据的边缘,获取待恢复投影数据的边界位置;然后引入权重,对待恢复点的邻域点进行预插值,并利用全变分模型对缺失的投影数据进行恢复;最后依据扩充较完整的投影数据,经滤波反投影算法重建得到改善的局部图像。与直接利用截断投影数据重建图像中呈现的环形伪影相比较,仿真和临床数据的重建图像对比度增强,且较好地保留图像的边缘信息。Aiming at the feature that the projection data is truncated along the direction of the detector during local scanning,a fan-beam CT local reconstruction algorithm based on projection data restoration was proposed.The algorithm used the line process model to detect the edges of the truncated projection data,and obtained the boundary position information of the projection data to be restored;then introduced weights,preinterpolated the neighborhood points of the pixels to be restored,and used the total variation model to restore the missing projection data;Finally,based on the expanded more complete projection data,the improved local image was reconstructed by the filtered back projection algorithm.Compared with the ring artifacts present in the reconstructed image directly using the truncated projection data,the contrast of the reconstructed image of simulation and clinical data is enhanced,and the edge information of the image is better preserved.

关 键 词：局部重建 环状伪影 投影数据恢复 边缘检测 全变分模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]