头颈部肿瘤分子生物纹理分析与生物靶区自适应勾画  

Molecular biological texture analysis and adaptive delineation of biological target volumes corresponding to head and neck tumors

作　　者：刘国才[1] 余志浩[1] 朱苏雨[2] 莫逸[3] 胡炳强[2] 张九堂[2] 阳维力[1] 吴海燕[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082 [2]中南大学湘雅医学院附属湖南省肿瘤医院放疗科,湖南长沙410013 [3]中南大学湘雅医学院附属湖南省肿瘤医院PET/CT中心,湖南长沙410013

出　　处：《中国医学影像技术》2013年第1期115-120,共6页Chinese Journal of Medical Imaging Technology

基　　金：国家自然科学基金(61271382;60835004;61072121;60872130)

摘　　要：目的探讨头颈部肿瘤分子生物纹理分析和自适应生物靶区(BTV)勾画方法。方法根据肿瘤PET图像的标准摄取值(SUV)及其分子生物"方差"(VAR)纹理特征,提出一种改进的两阶段自适应三维体生长肿瘤BTV勾画方法:首先根据PET图像灰度共生矩阵,提取肿瘤分子生物VAR纹理特征;之后联合肿瘤生物VAR纹理特征,改进先前提出的两阶段自适应三维体生长方法,并进行头颈部肿瘤BTV的自适应勾画。结果鼻咽癌临床PET影像试验和应用结果表明,改进的两阶段自适应三维体生长方法能自适应地勾画鼻咽癌BTV,结果正确合理。结论改进的两阶段自适应三维体生长方法能够更精确地自适应勾画头颈部肿瘤。Objective To assess molecular biological texture analysis and adaptive delineation of biological target volume （BTV） corresponding to head and neck tumors. Methods An improved two-stage adaptive three-dimensional volume-growing method for delineation of biological target volume was proposed based on the positron emission tomography （PET） standardized uptake value （SUV） image of tumors and features of biological texture namely Variance （VAR）. Firstly, molecular biological textures were extracted based on the gray-level co-occurrence matrix （GLCM） of PET SUV image of tumors. Then the extracted biological texture, namely VAR and the SUV features were simultaneously used to automatically delineate the target volume via an improved two-stage adaptive region-growing algorithm. Results Clinical application of nasopharyngeal carcinoma imaging were tested and evaluated by experienced radiation oncologists. BTV of nasopharyngeal carcinomas could be reasonably and adaptively delineated by the proposed method. Conclusion This proposed method can accurately and precisely delineate BTV corresponding to head and neck tumors.

关 键 词：分子生物纹理分析 自适应区域生长 生物靶区勾画 医学图像分割 头颈部肿瘤 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]