应用小波特征提取肺癌组织FTIR的支持向量机分类方法研究  

Study on the Methods of Wavelet Feature Extraction and SVM Classification of FTIR Lung Cancer Data

作　　者：程存归[1] 田玉梅[1] 金文英[2] 

机构地区：[1]浙江师范大学化学系浙江省固体表面反应化学重点实验室,金华321004 [2]义乌工商职业技术学院计算机工程系,义乌322000

出　　处：《化学学报》2007年第22期2539-2543,共5页Acta Chimica Sinica

摘　　要：提出了一种新的基于傅里叶变换红外光谱(Fourier Transform Infrared Spectroscopy,FTIR)的小波特征提取与支持向量机(SVM)分类方法以提高FTIR对早期肺癌的诊断准确率.对肺正常组织、早期肺癌及进展期肺癌组织的FTIR,利用连续小波(CW)多分辨率分析法提取9个特征量,支持向量机把其分为正常组与非正常组(包括早期肺癌和进展期肺癌),对正常组织、早期肺癌和进展期肺癌的识别,多项式核函数和径向基函数的识别准确率最高.多项式核函数对正常组织、早期肺癌和进展期肺癌的识别准确率分别为100%,95%及100%;径向基函数分别为100%,95%和100%.实验结果表明FTIR-CW-SVM模式分类方法对正常肺癌组织、早期肺癌及进展肺癌的识别具有较好的可行性.In order to improve the accuracy to earlier stage lung cancer diagnose rate with FTIR, a novel method of extraction of FTIR feature using wavelet analysis and classification using the support vector machine （SVM） was developed. To the FTIR of normal lung tissues, early carcinoma and advanced lung cancer, 9 feature variants were extracted with continuous wavelet （CW） analysis. With SVM, all spectra were classified into two categories： normal and abnormal ones, which included early lung cancer and advanced lung cancer. The accurate rates of poly and RBF kernel was high in all kernels. The accurate rates of poly kernel in normal, early lung cancer and advanced cancer were 100%, 95% and 100%, respectively and those of RBF kernel in normal, early lung cancer and advanced cancer were 100%, 95% and 100%, respectively. The research result shows the feasibility of establishing the models with an FTIR-CW-SVM method to identify normal lung tissue, early lung cancer and advanced lung cancer.

关 键 词：傅里叶变换红外光谱 小波特征提取 支持向量机 早期肺癌 

分 类 号：R734.2[医药卫生—肿瘤]