食源性致病菌近红外光谱改进贝叶斯判别模型的建立

Improved Bayes Discriminant Analysis on Near Infrared Spectroscopy of Food Pathogens

采用傅里叶变换近红外光谱仪结合化学计量学方法对大肠埃希氏菌O157∶H7、单增李斯特菌、金黄色葡萄球菌3种典型食源性致病菌进行鉴别研究。光谱数据经矢量归一化等预处理后,选取6 000~4 000 cm^(-1)波数范围具有菌株特性的谱图作为鉴别分析的研究对象,进行主成分分析(Principal component analysis,PCA)。利用PCA提取的前2个主成分作为贝叶斯判别(Bayes discriminant analysis,BDA)模型的最优判别因子,建立改进贝叶斯判别模型。该改进模型对训练样本和未知样本的判别正确率均达到100%。结果表明,改进BDA模型比BDA模型在食源性致病菌的判别精确度上有较大提高,基于近红外光谱的改进BDA模型可以作为一种准确、有效的食源性致病菌快速鉴别方法。

Three typical food pathogens that Escherichia coli 0157 were distinguished by Fourier transform near-infrared spectroscopy ： H7, Listeria monocytogenes and Staphylococcus aureus combined with stoichiometric method. Principal component analysis （PCA） was applied to the spectrum of bacteria strain peculiarity with range of 6 000-4 000 cm 1 which was selected as the research object of discriminatory analysis after vector normalization pretreatment for spectroscopic data. The first two principal components extracted by PCA were used as the optimal discrimination factors of Bayes discriminant analysis （BDA） model, and the improved BDA model were established. There are 100% accuracy that the improved BDA model identifies the training samples and the unknown samples. The results indicated that the improved BDA model has a considerable improvement compared with the single BDA method on the identification accuracy of food pathogens, and the improved BDA model based on Fourier transform near-infrared spectroscopy could be used as an accurate and efficient method for rapid identification of food pathogens.