基于拉曼光谱技术对丙型肝炎患者血清的检测

Detection of serum in patients with hepatitis C based on Raman spectroscopy

目的利用血清拉曼光谱建立快速诊断HCV感染的诊断模型。方法将HCV感染者血清(374例)与非HCV感染者血清(707例)吸至毛细管内于拉曼光谱仪进行点测量,获得拉曼光谱原始数据后,使用自适应迭代重加权惩罚最小二乘法扣除荧光背景,并采用origin2018软件寻找拉曼光谱数据之间的差异,构建airPLS-PLS-SVM统计模型。结果 HCV感染者血清与非HCV感染者拉曼光谱数据于442、509、621、643、 755 、828、853、876、925、 1 002 、 1 031 、1 047、1 210、1 330、1 449、1 555、1 670 cm^-1 位移上存在差异,于1 002、 1 157 、 1 515 cm^-1 位移处强度差异明显,绘制受试者工作特征曲线评估模型效果,得到特异度为97.424 9%,灵敏度为93.406 6%,正确率为96.296 3%。结论血清拉曼光谱技术结合airPLS-PLS-SVM统计模型能够很好地区分HCV感染与非HCV感染血清,有望成为一种快捷、简便的HCV感染诊断工具。

Objective To analyze the serum Raman spectra of the subjects and to establish a diagnostic model for the rapid diagnosis of HCV infection in the serum.Methods The serum of persons infected HCV (n=374) and the serum (n=707) of persons without HCV were sucked into capillary tube for point measurement in Raman spectrometer,and the original data of Raman spectroscopy were obtained,and the fluorescence background was deducted by using adaptive iterative weight weighted penalty least squares method.The origin2018 software was used to find the difference between Raman spectral data,and the statistical model of airPLS-PLS-SVM was constructed.Results Raman spectroscopy data of serum between non-HCV infected persons and HCV infected persons were different at 442,509,621,643,755,828,853,876,925,1 002,1 031,1 047,1 210,1 330,1 449,1 555,1 670 cm^-1 displacement,the intensity difference was obvious at 1 002,1 157,1 515 cm^-1 displacement,and the effect of receiver operator characteristics analysis curve evaluation model was obtained and showed that specificity was 97.424 9%,sensitivity was 93.406 6%,correct rate was 96.296 3%.Conclusion Serum Raman spectroscopy combined with airPLS-PLS-SVM statistical model can be a good distinction between HCV patients and non-HCV infected persons,which is expected to be a quick and easy diagnostic tool for HCV infection.