独立分量分析在伤口感染监测电子鼻技术中的应用

Independent Component Analysis for Wound Infection Using Electronic Nose Technology

针对传统的伤口感染诊断方法耗时长,操作复杂等问题,提出了一种基于电子鼻和独立分量分析(ICA)的方法来检测常见的伤口感染病原茵。该电子鼻的传感器阵列由6个金属氧化物半导体传感器组成,分别对七种常见病原菌产生响应,然后利用RBF神经网络对经ICA预处理后的数据进行识别。结果表明,ICA对气体传感器阵列测量数据进行预处理,可以简化神经网络的结构,减少计算量,并能提高伤口感染病原茵识别的准确率。

A method based on the electronic nose（e-nose） and independent component analysis（ICA） is presented to solve the time-consuming and complicated operation which appeared in traditional diagnosis method of wound infection. The gas sensor array of this e-nose consists of six metal oxide semiconductor sensors, which respond to the seven common pathogens in wound infection. The RBF （Radius Basis Function） neural network is used for pattern recognition after pre-processing of the ICA. The results show that preprocessing of the gas sensor array measurement data by the ICA can simplify the structure of neural network, with the computation complexity reduced and the recognition accuracy of the wound infection pathogens increased.