项目名称: 基于低辐射双能谱CT实现多种基础物质分解的重建算法研究
项目编号: No.61501292
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 无线电电子学、电信技术
项目作者: 龙泳
作者单位: 上海交通大学
项目金额: 21万元
中文摘要: 多能谱CT能够区分多种基础物质,为研究人体不同组织的性能和物体的不同材料成分提供特征鉴定和定量分析。基于模型的图像重建(MBIR)技术已被证明可以在常规单能谱CT上大幅度降低辐射剂量并提供高质量、高准确度的图像。但是,BMIR的主要缺陷就是运算处理时间太长。本课题将采用基于泊松噪声模型的最大惩罚似然估计,基于压缩感知原理的正则化方法和基于交替方向乘子法(ADMM)的优化算法,研究在双能谱CT上实现低辐射的准确的多种基础物质分解的快速MBIR重建方法。本课题提出快速且准确的同时三维正反向投影多种基础物质的方法,来突破MBIR的主要计算瓶颈。双能谱CT是目前唯一的多能谱CT产品。本课题提出的MBIR重建方法,在临床、工业以及国防安全领域有广阔的应用前景,特别是在降低双能谱CT对病人的电离辐射危害,提高肿瘤及其他疾病的早期诊断和治疗的有效性,提高检测爆炸物和毒品等危险物品的准确性方面有重要意义。
中文关键词: 多能CT;图像重建;反演算法;XCT成像
英文摘要: Spectral CT provides information on material characterization and quantification because of its ability to separate multiple basis materials. Model-based image reconstruction (MBIR) has been demonstrated on regular single-energy CT for improving the ability to produce high-quality and accurate images, while substantially reducing dose. However, MBIR operates with one major drawback of the much longer computation time. Using penalized-likelihood method based on Poisson noise distribution, regularization method based on compressed sensing theory and optimization method based on alternating direction method of multipliers (ADMM), we propose to develop a fast MBIR method to accurately decompose multiple basis materials from low-dose dual-energy CT (DECT). We propose a fast and accurate three-dimensional method to simultaneously forward and back project multiple basis materials, which is the primary computational bottleneck of MBIR. DECT is currently the only commercially available version of spectral CT. The proposed DECT based MBIR method has promising applications in clinic, industry and homeland security. It will reduce ionizing radiation damage to patients, improve the efficacy of early diagnosis and treatment of cancer and other diseases using DECT, and improve the accuracy of detecting explosives, narcotics and other dangerous objects.
英文关键词: Spectral CT;Image reconstruction;inverse problem;XCT imaging