计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (12): 238-240.

• 工程与应用 • 上一篇    下一篇

一种激光诱导荧光光谱特征提取新方法

王双维,樊晓平,廖志芳   

  1. 中南大学 信息科学与工程学院,长沙 410083
  • 收稿日期:2007-07-31 修回日期:2007-10-19 出版日期:2008-04-21 发布日期:2008-04-21
  • 通讯作者: 王双维

New feature extraction method for laser-induced fluorescence spectra

WANG Shuang-wei,FAN Xiao-ping,LIAO Zhi-fang   

  1. School of Information Science and Engineering,Central South University,Changsha 410083,China
  • Received:2007-07-31 Revised:2007-10-19 Online:2008-04-21 Published:2008-04-21
  • Contact: WANG Shuang-wei

摘要: 利用主成分分析(Principal Component Analysis,PCA)和Fisher线性判别分析(Fisher Linear Discriminative Analysis,FLDA)方法相结合提取特征,提出了一种荧光光谱特征提取新方法——PCA_FLDA。实验证明,新方法提高了激光诱导自体荧光光谱对早期结肠癌的诊断精度。对预处理后的240条荧光光谱,利用PCA_FLDA算法提取了50个特征变量,利用支持向量机将其分为正常组织和癌变组织,分类敏感性、特异性和准确率可分别达到97.5%、95.12%和96.25%。

关键词: 激光诱导荧光光谱, 特征提取, PCA, FLDA, 支持向量机

Abstract: Combined with Principal Component Analysis (PCA) and Fisher Linear Discriminative Analysis (FLDA) to extract features,a novel fluorescence spectra feature extraction algorithm named PCA_FLDA is presented in this paper.The experiment results show that this method can improve the diagnostic rate of earlier stage colonic cancer with laser-induced fluorescence spectra.After preprocessing the collected 240 spectra,50 feature variants were extracted with PCA_FLDA.By means of the support vector machine,all spectra are classified into two categories as the normal or the abnormal one.The sensitivity,specificity and discriminating accuracy are reached to 97.5%,95.12% and 96.25%,respectively.

Key words: laser-induced fluorescence spectra, feature extraction, Principal Component Analysis(PCA), Fisher Linear Discriminative Analysis(FLDA), Support Vector Machine(SVM)