腮腺良性肿瘤拉曼光谱特征及诊断模型研究

Research on Raman spectral characters and diagnostic discriminating model of parotid benign tumors

目的研究腮腺良性肿瘤拉曼光谱(Raman spectra)特征,并结合统计学方法建立诊断分类模型。方法收集腮腺正常组织、腮腺多形性腺瘤和Warthin瘤组织样本各20例,以785nm波长近红外激发光对样本进行拉曼光谱扫描,分析不同组织光谱的特征,应用主成分分析法(principle component analysis,PCA)与线性判别函数分析(liner discriminantan alysis,LDA)相结合的方法建立诊断分类模型。结果比较不同组织的平均拉曼光谱,腮腺多形性腺瘤较腮腺正常组织在蛋白质、核酸和脂类的峰位谱峰增强,Warthin瘤在蛋白质和核酸峰位的谱峰较正常增高,而脂类峰位谱峰明显降低。通过PCA-LDA建立诊断分类模型,其总体分类准确率达90%以上。结论腮腺正常组织、多形性腺瘤与Warthin瘤组织的拉曼光谱存在差异,通过PCA-LDA建立诊断分类模型,可以鉴别区分3者。

OBJECTIVE To study the Raman spectral characters of parotid benign tumors and establish a diagnostic discriminating model.METHODS A total of 60 samples were collected from normal parotid gland, pleomorphic adenoma and Warthin＇s tumor.Raman spectra were gained by Raman microscope with a 785nm excitation.PCA-LDA was used to establish the diagnostic model for discrimination of these spectra. RESULTS There were significant differences among the mean Raman spectra of these 3 kinds of samples. The mean spectrum of pleomorphic adenoma showed strong peaks contributed to excess nucleic acids,proteins and lipids,and the mean spectrum of Warthin＇s Tumor showed strong peaks contributed to nucleic acids and proteins but weak peaks contributed to lipids.Through the discrimination model established by PCA-LDA,the total accuracy reached over 90%.CONCLUSION There were differences existing in the Raman spectra of different tissues,and the diagnostic model could discriminate the one from the others with a high accuracy.