基于支持向量机法的近红外光谱鉴别耐甲氧西林金葡菌和甲氧西林敏感金葡菌的研究

发布时间：2018-09-03 08:48

【摘要】：目的探讨使用近红外光谱结合支持向量机法区分耐甲氧西林金葡菌(MR SA)和甲氧西林敏感金葡菌(MSSA)的可行性。方法制作MRSA和MSSA的浓度标准曲线。扩增待测细菌,并根据公式制备相同浓度菌液。采集菌液样本近红外光谱数据,并对数据进行一阶求导、平滑去噪、归一化和基线校正等预处理。根据两种细菌光谱曲线的相关性,对900~2200nm波段数据进行主成分分析。依据累计贡献率结果,选择前三个主成分作为支持向量机的输入向量,分别使用线性、多项式、径向基3种核函数进行建模,比较不同模型区分MSSA和MRSA的准确性。结果 MRSA和MSSA预处理后的光谱曲线相关系数为1.000,两者高度相似。使用主成分处理并采用3种支持向量机核函数建模后,模型的训练和测试准确率均高于95%,其中采用径向基核函数分类结果最好,训练准确率为99.72%±0.21%,测试准确率为99.47%±0.00%。结论使用近红外光谱结合支持向量机的分析方法具有精确区分MRSA和MSSA的能力。
[Abstract]:Objective to study the feasibility of using near infrared spectroscopy (NIR) combined with support vector machine (SVM) to distinguish methicillin-resistant Staphylococcus aureus (MR SA) from methicillin-sensitive Staphylococcus aureus (MSSA). Methods the standard concentration curves of MRSA and MSSA were made. The bacteria to be tested were amplified and the same concentration of bacteria was prepared according to the formula. The near infrared spectrum data of bacterial liquid samples were collected, and the data were preprocessed by first order derivation, smooth denoising, normalization and baseline correction. According to the correlation of two bacterial spectral curves, principal component analysis (PCA) of 900~2200nm band data was carried out. According to the result of cumulative contribution rate, the first three principal components are selected as input vectors of support vector machine, and three kernel functions, linear, polynomial and radial basis function, are used to model and compare the accuracy of different models to distinguish MSSA from MRSA. Results the correlation coefficient of spectral curve between MRSA and MSSA was 1.000, and they were highly similar. The training and testing accuracy of the model is higher than that of 95% by using principal component processing and using three kinds of kernel function of support vector machine. The classification result of radial basis kernel function is the best, the training accuracy is 99.72% 卤0.21, and the test accuracy is 99.47% 卤0.00. Conclusion the method of near infrared spectroscopy combined with support vector machine has the ability to distinguish MRSA from MSSA accurately.
【作者单位】：第三军医大学西南医院全军烧伤研究所创伤、烧伤与复合伤国家重点实验室;重庆大学通信工程学院通信工程系;中国工程物理研究院流体物理研究所;
【分类号】：R446.5

【相似文献】