基于Mean Shift和插值图像修复算法的CT图像金属伪影消除方法

发布时间：2018-05-20 11:35

本文选题：金属伪影校正 + Mean　；参考：《电子学报》2017年08期

【摘要】：近年来CT成像技术在临床医学中广泛应用,但当病人体内含有金属移植物时,由于射线硬化等原因很可能在金属物体周围产生亮暗伪影,降低图像质量,影响诊断的准确性.为了消除CT图像中的金属伪影,本文提出基于Mean Shift和插值图像修复的算法,基本流程为用自适应Mean Shift算法预处理CT图像,平滑噪声和轻度伪影,用简化的Mean Shift算法快速精确分割金属物体,由修复组织信息的插值图像生成先验图像,用先验图像的投影数据替换原投影数据得到校正后的CT图像.经过对比实验,文中算法在去除金属伪影的同时,能够保护原有CT图像的组织结构,取得了更好的处理效果.
[Abstract]:In recent years, CT imaging technology has been widely used in clinical medicine, but when patients have metal grafts in their bodies, it is very possible to produce bright and dark artifacts around metal objects due to ray hardening, which will reduce the quality of images and affect the accuracy of diagnosis. In order to eliminate metal artifacts in CT images, an algorithm based on Mean Shift and interpolation image restoration is proposed in this paper. The basic flow is to preprocess CT images with adaptive Mean Shift algorithm to smooth noise and slight artifacts. A simplified Mean Shift algorithm is used to segment metal objects quickly and accurately. A priori image is generated from the interpolated image of repairing tissue information, and the corrected CT image is obtained by replacing the original projection data with the projection data of the priori image. Through the contrast experiment, the algorithm not only removes the metal artifacts, but also can protect the structure of the original CT image and obtain better processing effect.
【作者单位】：吉首大学信息科学与工程学院;中南大学信息科学与工程学院;
【基金】：国家自然科学基金(No.61562029,No.61363073,No.61262032) 湖南省自然科学基金(No.2015JJ3100) 吉首大学学成返校取得博士学位、学历人员科研资助经费研究项目(吉大人通[2014]39号)
【分类号】：R814.42;TP391.41

【相似文献】