基于单细菌共焦拉曼光谱的细菌快速检测

Rapid identification of five species of bacteria based on confocal Raman spectroscopy of single bacterial cells

摘要:

【背景】目前利用共焦拉曼光谱技术进行成像和成分鉴别方面的研究较多，但如何快速检测与鉴别多种细菌方面的研究较少。【目的】基于共焦拉曼光谱技术，建立一种在单细菌水平上实现病原微生物快速分类鉴定的方法。【方法】以大肠杆菌为研究对象，利用共焦拉曼光谱技术在单细菌水平上进行了激发波长的优化试验，并研究了大肠杆菌存放时间对单细菌拉曼光谱信息的影响。同时，对白色葡萄球菌、大肠杆菌、金黄色葡萄球菌、沙门氏菌和铜绿假单胞菌进行了共焦拉曼光谱测试，并对5种细菌进行单细菌拉曼光谱的归属分析，设计共焦拉曼光谱技术结合支持向量机(support vector machine，SVM)模型学习算法，进行了5种细菌的快速分类鉴别。【结果】对于单细菌拉曼光谱探测，532、633和785 nm这3种常见的拉曼探测波长中，532 nm具有更好的激发效率和光谱信噪比。结合SVM模型对5种细菌的识别分类，SVM模型的灵敏度和特异性达到了96.00%以上，整体准确率为98.25%。不同存放时间下大肠杆菌拉曼光谱的重复性和稳定性都很好，且SVM模型匹配率均在90.00%以上。【结论】单细菌拉曼光谱结合SVM模型可对5种细菌进行快速、准确的分类，不同存放时间对大肠杆菌拉曼光谱的归类识别几乎无影响。

关键词:共焦拉曼光谱技术;单细菌;分类鉴定;线性支持向量机模型

Abstract:

[Background] Despite the extensive research on the application of confocal Raman spectroscopy in imaging and component identification, there are few studies focusing on the rapid detection and identification of a variety of bacteria. [Objective] A method for rapid classification and identification of single bacterial cells was established with confocal Raman spectroscopy. [Methods] We optimized the excitation wavelength of confocal Raman spectroscopy by using the single cells of Escherichia coli, and studied the influence of storage time on the Raman spectrum of E. coli. Furthermore, we performed confocal Raman spectroscopy tests for Staphylococcus albus, Escherichia coli, Staphylococcus aureus, Salmonella, and Pseudomonas aeruginosa. We then analyzed the Raman spectra of the 5 bacterial species to design a rapid identification method combining confocal Raman spectroscopy with support vector machine (SVM) model for the 5 bacterial species. [Results] Among the three common Raman detection wavelengths of 532, 633, and 785 nm, 532 nm had the best excitation efficiency and spectral signal-to-noise ratio for the identification of single bacterial cells. The SVM model showed the sensitivity and specificity above 96.00% and the overall accuracy rate of 98.25% for the identification of these bacteria. Moreover, the Raman spectra showed good repeatability and stability for the E. coli stored for different time, and the matching rate of SVM model was above 90.00%. [Conclusion] Single-cell Raman spectroscopy combined with SVM model can quickly and accurately classify the five bacterial species, and different storage time has little effect on the identification of E. coli based on Raman spectra.

Key words:confocal Raman spectroscopy;single bacterial cells;classification and identification;support vector machine