利用PCA-LDA和参数-CART方法对直肠癌血清拉曼光谱的分析  

Raman Spectral Analysis of Rectum Cancer by Serum Using PCA-LDA and Parameter-CART Methods

作　　者：李晓舟[1] 孙宝明 杨天月[1] 金莉莉[3] 王德力[1] 于壮[1] 

机构地区：[1]沈阳理工大学理学院,辽宁沈阳110159 [2]沈阳永原科技开发有限公司,辽宁沈阳110006 [3]辽宁大学生命科学学院,辽宁沈阳110036

出　　处：《激光生物学报》2015年第1期73-76,共4页Acta Laser Biology Sinica

摘　　要：本文利用拉曼光谱技术检测由直肠癌的发生导致的血清成分的变化。比较了直肠癌患者和对照组之间血清拉曼光谱的差异,并对术后直肠癌患者血清拉曼光谱的变化也进行了比较,以监测术后治疗效果。结果表明在某些波数位置不同组的拉曼峰有统计学意义的变化,这些变化反应了血清中相应的生物物质的改变。之后,主成分分析(PCA)及峰强比参数这两种方法被用于原始拉曼光谱的特征变量的提取。将线性判别分析(LDA)和分类回归树(CART)两种判别分析法用于特征变量的判别分析。PCA-LDA和参数-CART方法的诊断准确率分别为87%和90%。Raman spectroscopy technique was used in this paper to detect the changes of the components in serum ac-company with the occurrence of rectum cancer.The differences in Raman spectroscopy of serum between rectum cancer and the healthy controls were examined.Post-operative rectum cancer patients were also participated in the comparison to monitor the effects of the operation.The results showed that there are significant variations at certain wavenumbers which are predications of the alteration of corresponding biological substances.Then,principal component analysis （PCA）and parameters of intensity ratios were used on the original Raman spectra for the extraction of featured variables.Linear dis-criminant analysis （LDA）and classification and regression tree （CART）were used on the featured variables for the dis-crimination analysis.Accuracies of 87% and 90% were got for separate PCA-LDA and Parameter-CART.

关 键 词：拉曼光谱 血清 直肠癌 诊断 

分 类 号：O657.37[理学—分析化学] R730.57[理学—化学]