使用激光共聚焦显微拉曼光谱仪测取膀胱肿瘤和正常膀胱组织的拉曼特征谱, 应用主成分分析/支持向量机(principal component analysis, PCA/support vector machines, SVM)分类器对数据进行判别分析, 最后使用弃一交叉验证法(leave-one-out cross validation, LOOCV)测试判别结果的准确度。 结果发现膀胱肿瘤组织与正常膀胱组织的拉曼光谱存在明显差异, 肿瘤组织在782和1 583 cm-1等核酸特征谱带处峰高显著增强, 而正常组织在1 061, 1 295, 2 849, 2 881 cm-1等蛋白质和脂质特征谱带处峰高显著增强。 PCA/SVM可良好区分膀胱肿瘤组织和正常膀胱组织的拉曼光谱, LOOCV测试分类器显示肿瘤诊断的敏感度86.7%、 特异度87.5%、 阳性预测值92.9%、 阴性预测值77.8%。 由此得出结论: 拉曼光谱可以良好诊断膀胱肿瘤的体外组织, 展现了优越的临床应用前景。

The scope of this research lies in diagnosis of bladder cancer through Raman spectra. The spectra of bladder cancer and normal bladder were measured by using laser confocal Raman micro-spectroscopy. Principal component analysis/support vector machines was applied to the spectral dataset to construct diagnostic algorithms, then to detect the accuracy of these algorithms to determine histological diagnosis by leave-one-out cross validation from its Raman spectrum. It was showed that the peak intensity of nucleic acid (782, 1 583 cm-1) in bladder cancer and protein (1 061, 1 295, 2 849, 2 881 cm-1) in normal bladder increased significantly. Additionally, Principal component analysis (PCA) and support vector machines (SVM) provided an effective tool for differentiating the bladder cancer from normal bladder tissue. Excellent sensitivity (86.7%), specificity (87.5%), positive predictive value (92.9%), and negative predictive value (72.8%) for the diagnosis of bladder cancer were obtained by leave-one-out cross validation. It was concluded that Raman spectroscopy can be used to accurately identify bladder cancer in vitro, and it suggests the promising potential application of PCA/SVM-based Raman spectroscopy for the diagnosis of bladder cancer.