基于在体植入电极的生物嗅觉传感系统设计及其气味识别  

Design of Biomimetic Olfactory Sensing System by Implanted Microelectrode Array and Its Application in Odor Recognition

作　　者：周俊[1] 董琪[1] 庄柳静[1] 吴春生[1] 刘清君[1] 王平[1] 

机构地区：[1]浙江大学生物医学工程与仪器科学学院,生物传感器国家专业实验室,生物医学工程教育部重点实验室,杭州310027

出　　处：《传感技术学报》2012年第8期1023-1028,共6页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(81027003)

摘　　要：目前生物嗅觉系统在气味识别方面相比于化学传感器阵列构成的电子鼻系统具有更高的灵敏度、特异性和响应速度。为了探讨生物嗅觉传感系统气味识别的可行性,构建了基于微电极阵列传感器植入大鼠嗅球构成的嗅觉传感系统,研究记录了浓度为10 mM的异丁醇、苯甲醚、香芹酮和柠檬醛4种气味刺激引起的嗅球僧帽层低频场电位信号,采用多窗谱估计算法和移动窗技术结合得到随时间分布的功率谱密度图。实验结果发现气味刺激后信号功率谱能量较多分布在gamma频段(40 Hz～120 Hz)。使用K最邻近分类方法对120组数据进行分类识别,4种气味分类正确率达到77.4%。实验结果表明该嗅觉传感系统结合多窗谱估计时频图与K最邻近分类算法可以初步实现气味识别。Mammalian olfactory system has merits of higher sensitivity,selectivity and faster response than current electronic nose system based on chemical sensor array in odor recognition.In order to discuss the feasibility of odor discrimination by odor sensing system,the microelectrode array was implanted into olfactory bulb of rats to record local field potential from mitral cell layer after stimulated by four odors including isobutanol,anisole,carvone and citral.The spectrograms using multitaper spectrum estimation and time-shift windows show relative high energy distribution in gamma band（40 Hz～120 Hz） after odor stimulus.A data set consists of 120 odor-induced spectrograms was used for odor classification by K-nearest neighbor method and the classification accuracy reached 77.4%.The result indicates that the olfaction based odor sensing system is capable of discriminating different odors.

关 键 词：嗅觉传感系统 气味识别 嗅球场电位 多窗谱估计 K最邻近分类算法 

分 类 号：TP212.3[自动化与计算机技术—检测技术与自动化装置]