复杂性测度特征在肺部HRCT图像分析中的应用  

Application of Complexity Measurements to Lung HRCT Image Analysis

作　　者：陈旭[1] 庄天戈[1] 滑炎卿[2] 

机构地区：[1]上海交通大学生物医学工程系,上海200030 [2]上海华东医院放射科,上海200040

出　　处：《上海交通大学学报》2003年第2期224-227,共4页Journal of Shanghai Jiaotong University

基　　金：上海市科技发展基金资助项目 ( 0 0 4 1 1 90 1 1 )

摘　　要：研究了分形维和 Lempel- Ziv(LZ)复杂性两类基于视觉复杂性的图像特征在自动区分高分辨率 CT(HRCT)上磨玻璃影 (GGO)与正常区域的表现 .研究样本包括 86个 1 5× 1 5大小的矩形感兴趣区 (ROI) ,其中 44个正常 ,42个 GGO.将从这些 ROI中提取的分形维特征和 LZ复杂性特征作为输入对线性分类器训练并对其分类性能进行评估 .结果表明 ,若将两类特征单独作为分类器的输入 ,相应的 ROC曲线下面积分别为 0 .837和 0 .90 3;当用回代法训练和测试分类器时 ,分别有 75.6%和 79.1 %的 ROI被正确分类 ,而用刀切法时 ,ROI被正确分类的比率相同 .若将两类特征的组合作为分类器输入 ,相应的 ROC曲线下面积提高到 0 .969,而总的分类正确率亦达 91 .9%This paper investigated two image features based on visual complexity measurements:the fractal dimension (FD) and the Lempel Ziv complexity (LZC), and evaluated their performance in differentiating GGOs from normal areas on lung HRCT images. The database of this study contains 86 rectangular ROIs (44 Normal, 42 GGO) of 15×15 pixels. The features of FD and LCZ extracted from these ROIs were input to a linear classifier to predict their classification. When the two features were used individually, they respectively yielded areas under the ROC curve (AUC) of 0.837 and 0.903; 75.6%/79.1% of ROIs were correctly classified when training and testing in a re substitution as well as a jackknife procedure. On condition that both features were input to the classifier, an AUC of 0.969 was achieved; meanwhile the overall accuracy increased up to 91.9%. The promising results demonstrate the FD and LZC's potential in GGO discrimination.

关 键 词：图像分析 视觉复杂性 分形维 Lempel-Ziv复杂性 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] R318[自动化与计算机技术—计算机科学与技术]