基于规则及多特征跟踪的肺结节的智能检测方法  

Intelligent Detection of Lung Nodules based on Rules and Multi-feature Tracking

作　　者：王立非 李彬[2] 田联房[2] 刘思伟[2] 谭利[2] 张婧[2] 

机构地区：[1]广州中医药大学祈福医院影像中心,广州511495 [2]华南理工大学自动化科学与工程学院,广州510640

出　　处：《生物医学工程研究》2010年第2期79-83,105,共6页Journal Of Biomedical Engineering Research

基　　金：中国博士后基金资助项目(20090450866);广东省自然科学基金资助项目(8451064101000631);广州市番禺区科技攻关项目(2009-Z-108-1);教育部博士点基金资助项目(200805610018);广东省教育部产学研结合项目(2009B090300057)

摘　　要：肺结节的智能识别对肺癌的诊断至关重要。为了在大量的肺部CT图片中准确智能识别肺结节,我们研究了一个基于规则及多特征跟踪的肺结节计算机辅助检测方法。其中,采用活动轮廓模型的分割方法实现候选肺结节分割,采用基于规则的决策方法以及多特征跟踪方法实现肺结节分类。实验证明,该肺结节的智能检测方法满足肺结节计算机辅助诊断的要求。Intelligent recognition of lung nodules which are distinct signs of lung cancer is important for diagnosis of lung cancer.In order to accurately identify lung nodules in a large amount of thorax CT images,a method of Computer-Aided Detection（MCAD） for lung nodules based on rules and multi-feature tracking was proposed in this paper.In this proposed method,candidate pulmonary nodules were segmented by GVF snake model;lung nodules were classified based on rules and multi-feature tracking method.Experiments demonstrate that the MCAD provide a good means for computer-aided diagnosis of lung nodules and cancer.

关 键 词：肺结节 计算机辅助检测 分类 医学图像 识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] R319[自动化与计算机技术—计算机科学与技术]