论文部分内容阅读
径向基函数(RBF)网络被用于根据基质辅助激光解吸/电离飞行时间质谱(MALDI-TOF-MS)对细菌的分类辨识。为了加速网络训练和减少网络的复杂性,本文采用小波变换对原始质谱数据进行压缩,将原来的13828个数据点压缩至328个,且保持了原来的特征谱峰。本文研究了在不同培养时间(24、48和72小时)的5种细菌分类,并对RBF网络参数的影响做了详细地研究,为生物学研究提供了有用的信息。结果表明,约60％以上的细菌样本能够被正确地分类辨识。由于细菌培养的生物学影响因素复杂,因而进一步严格控制细菌的培养条件是改善细菌分类正确率的关键。
Radial Basis Function (RBF) networks were used to classify bacteria based on matrix-assisted laser desorption / ionization time-of-flight mass spectrometry (MALDI-TOF-MS). In order to speed up network training and reduce the complexity of the network, this paper uses wavelet transform to compress the original MS data, compressing the original 13828 data points to 328 and keeping the original characteristic peaks. In this paper, we studied five bacterial species at different incubation times (24, 48 and 72 hours) and studied the effects of RBF network parameters in detail, providing useful information for biological research. The results show that more than 60% of bacterial samples can be correctly classified. Due to the complicated biological influence of bacterial culture, further strict control of bacterial culture conditions is the key to improve the accuracy of bacterial classification.