肺部模糊区域的实质自动图像分割方法研究  

Research on the Automatic Image Segmentation method in Fuzzy Region of Lung Parenchyma

作　　者：曾羽琚[1,2] 陈明辉[3] 

机构地区：[1]长沙环境保护职业技术学院信息技术系,湖南长沙410007 [2]湖南大学信息科学与工程学院,湖南长沙410082 [3]湖南商学院,湖南长沙410205

出　　处：《计算机仿真》2014年第3期376-379,396,共5页Computer Simulation

摘　　要：研究CT序列图像肺部准确分割问题。在CT图像中,左右肺前后连接线极其狭窄,肺部边缘产生结节引起模糊,传统的高斯图像分割方法会造成像素连通域标记去除背景困难,细小空洞易发生分割遗漏及左右肺分离等问题。为了避免上述缺陷,提出自适应阀值控制的CT序列图像肺部自动分割方法。利用小波变换方法,去除肺部图像中的横纹噪声,提高图像的分辨率。设定自适应阀值控制,获取CT图像的自适应阀值,从而完成肺部CT序列图像的自动分割。实验结果表明,改进算法进行CT序列图像的肺部自动分割,能够提高分割的准确率、分割灵敏度和特异性。The accurate segmentation method for CT sequence image of lung parenchyma was studied in this pa- per. In the CT images, the back and forth connecting line of left and right lungs are extremely narrow, and the nod- ules generated in lung parenchyma edge will cause blurring. The traditional Gaussian image segmentation method can cause that background removal of pixel connected domain label is difficult and that tiny cavity region, are easy to oc- cur problems of omission segmentation and left and right lungs separation. In order to avoid these defects, this paper puts forward an automatic segmentation method for CT sequence image of lung parenchyma based on adaptive thresh- old control. Using the method of wavelet transform, the horizontal striped noise in the lung image can be removed to improve the image resolution. Setting adaptive threshold control, the adaptive threshold of CT image is obtained to complete the CT sequence image automatic segmentation of lung parenchyma. The Experimental results show that the improved algorithm for CT sequence image automatic segmentation of lung parenchyma can increase the accuracy, sensitivity and specificity of segmentation.

关 键 词：序列图像 自动分割 横纹噪声 临床医学 

分 类 号：F127[经济管理—世界经济]