摘要
CT广泛应用于临床诊断,但是当被扫描部位有金属植入物时,成像结果存在严重的金属伪影,影响医生诊断,因此对金属伪影进行校正具有重要的临床意义。本文首先介绍金属伪影产生的原因,接着梳理了传统的基于投影域数据修复及基于迭代的金属伪影去除方法,并主要关注了近年来兴起的深度学习技术在金属伪影去除领域的一些应用方法,最后对这些方法进行总结并对金属伪影去除的前景做出展望。
Computed tomography(CT)has been widely used in clinical diagnosis.However,when metal implants exist in a scanned target,there will be serious metal artifacts in the imaging result,which will heavily affect the doctor’s diagnosis.As a result,it is necessary to develop effective metal artifact reduction(MAR)methods.This paper firstly introduces the causes of metal artifacts,and then the traditional MAR methods,including projection domain data restoration and iterative reconstruction.Specially,with the rapid development of the emerging deep learning technologies in recent years,this paper focuses on the application of deep learning in MAR.Finally,we summarize current methods and give some potential solutions for MAR.
作者
汪涛
夏文军
赵云松
张意
WANG Tao;XIA Wenjun;ZHAO Yunsong;ZHANG Yi(College of Computer Science,Sichuan University,Chengdu 610065,China;School of Mathematical Sciences,Capital Normal University,Beijing 100048,China)
出处
《中国体视学与图像分析》
2020年第3期207-223,共17页
Chinese Journal of Stereology and Image Analysis
基金
国家自然科学基金(No.61671312)
关键词
CT成像
金属伪影去除
深度学习
插值
迭代
CT imaging
metal artifact reduction
deep learning
interpolation
iterative reconstruction
作者简介
汪涛(1996-),男(汉),安徽人,硕士研究生。研究方向:医学图像处理;通信作者:张意,副教授。E-mail:yzhang@scu.edu.cn
