ITEAR算法:以纹理能量分析为基础的早期肿瘤图像检测方法  

ITEAR Algorithm:Early Stage Tumor Image Recognition Method Basing on Texture Energy Analysis

作　　者：何元烈[1] 陈萍[2] 陈仰纯[2] 战荫伟[1] 

机构地区：[1]广东工业大学计算机学院,广东广州510006 [2]广州医学院第一附属医院核医学科,广东广州510120

出　　处：《广州医学院学报》2009年第3期7-11,共5页Academic Journal of Guangzhou Medical College

基　　金：国家自然科学基金(30800274);广州市卫生局基金(2007-YB-159)

摘　　要：目的:实现计算机自动检测出CT图像中的早期肿瘤。方法:提出一个以纹理能量分析为基础的可以检测出CT图像中早期肿瘤的算法——ITEAR(image registration and texture energy analysis)。该算法首先用互信息量配准方法将PET图像和同一层的CT图像配准,并且利用自适应阈值法将PET图像中的感兴趣区(region of interest,ROI)检测出来,然后应用纹理能量分析方法检测出早期肿瘤的准确位置,接着运用边缘检测方法将其边界提取出来。结果:ITEAR算法能准确定位早期肿瘤,并且能检测出早期肿瘤的边界。结论:ITEAR算法可以令计算机自动制定三维适形放疗计划。Objective： Automatically identifying the early stage tumor in CT image by computer. Method：This paper develops the ITEAR （ Image Registration and Texture Energy Analysis） algorithm, which can accurately identify the early stage tumor in CT images. The ITEAR algorithm aligns the corresponding PET and CT images by the maximum mutual information registration method and uses the adaptive threshold segmentation algorithm to segment the region of interest （ROI） in PET images. The texture energy measurement method and edge detection processes are implemented to locate the accurate position of the early stage tumor and extract its contour. Results： It has been proved by experiments that The ITEAR algorithm can be used to accurately locate the early stage tumor in any part of human body. Conclusion： The ITEAR algorithm can be applied to develop automatic 3D conformal radiation therapy planning system.

关 键 词：肿瘤 CT图像 纹理 

分 类 号：R596[医药卫生—内科学]