Goodacre R, Rooney P J, Kell D B
Institute of Biological Sciences, University of Wales, Aberystwyth, UK.
J Antimicrob Chemother. 1998 Jan;41(1):27-34. doi: 10.1093/jac/41.1.27.
Curie-point pyrolysis mass spectra were obtained from 15 methicillin-resistant and 22 methicillin-susceptible Staphylococcus aureus strains. Cluster analysis showed that the major source of variation between the pyrolysis mass spectra resulted from the phage group of the bacteria, not their resistance or susceptibility to methicillin. By contrast, artificial neural networks could be trained to recognize those aspects of the pyrolysis mass spectra that differentiated methicillin-resistant from methicillin-sensitive strains. The trained neural network could then use pyrolysis mass spectral data to assess whether an unknown strain was resistant to methicillin. These results give the first demonstration that the combination of pyrolysis mass spectrometry with neural networks can provide a very rapid and accurate antibiotic susceptibility testing technique.
从15株耐甲氧西林金黄色葡萄球菌菌株和22株甲氧西林敏感金黄色葡萄球菌菌株中获得了居里点热解质谱。聚类分析表明，热解质谱之间的主要变异来源是细菌的噬菌体组，而非它们对甲氧西林的耐药性或敏感性。相比之下，可以训练人工神经网络来识别热解质谱中区分耐甲氧西林菌株和甲氧西林敏感菌株的那些方面。经过训练的神经网络随后可以利用热解质谱数据来评估未知菌株是否对甲氧西林耐药。这些结果首次证明，热解质谱与神经网络相结合可以提供一种非常快速且准确的抗生素敏感性检测技术。
