摘要
为探讨基于肺小结节CT图像灰度共生矩阵纹理特征的多水平模型,对北京友谊医院和宣武医院提供的185例2137张肺小结节CT图像提取4种灰度共生矩阵纹理特征。根据该类资料具有层次结构的特点,拟合多水平统计模型。结果表明,能量,熵和惯性矩,在患者水平上具有聚集性,且在良恶性肺小结节间的差异有统计学意义(P值均小于0.001),提示多水平模型可以灵活有效地处理肺小结节CT图像这类具有层次结构的数据,在一定程度上有利于早期肺癌的鉴别诊断。
To study the multilevel model base on texture features of CT image of small solitary pul- monary nodules, which were extracted by applying gray level co-occurrence matrix. Extracting 4 texture features based on gray level co-occurrence matrix from 2137 CT images of 185 patients supplied by friend hospital and XuanWu hospital of Beijing. According to the characteristic of hierarchical structure belonging to collected data, multilevel model was constructed. The result showed that three texture features, including energy, entropy and moment of inertia, own aggregating character on level of patient, and are significantly different （P 〈 0.001） in benign and malignant small solitary pulmonary nodules. The study suggested the multilevel model can process hierarchical structure data efficiently, such as CT images of small solitary pulmonary nodules, by which profit diagnosis of earlier period lung cancer to some extent.
出处
《数理统计与管理》
CSSCI
北大核心
2009年第4期756-760,共5页
Journal of Applied Statistics and Management
基金
北京市教育委员会科技发展计划面上基金(KM200610025014)资助
首都医科大学基础与临床基金课题(2006JL57)资助
关键词
纹理特征
肺小结节
多水平模型
CT图像
texture extraction, small pulmonary nodules, multilevel model, CT image
