摘要
丙酮可以作为糖尿病呼气检测的一种标志物。从呼气检测角度出发,基于电子鼻原理设计了糖尿病呼气检测分析仪。研究中基于金属氧化物半导体(Metal Oxide Semiconductor,MOS)气体传感器设计了传感器阵列,基于末潮呼气缓冲原理(Buffered end-tidal,BET)设计了呼气进样器,基于Matlab通过主成分分析法(Principal Component Analysis,PCA)和BP(Back-propagation)神经网络对呼气进行了模式识别。在模拟试验中,利用100份健康的呼气样本和100份模拟病患的呼气样本对系统的功能进行了验证。PCA图谱显示气体传感器阵列对两类不同的呼气样本的响应有着明显的区别,呼气进样器可实现系统的在线式进样,模式识别算法对两种样本的最终识别精度为91.67%。呼气分析仪的设计对于糖尿病的无入侵式检测有积极的意义。
Acetone could be regarded as a biomarker for diabetes.From the point of breath detection,a diabetes breath detection analyzer was designed based on the principle.A sensor array based on MOS sensors was designed in the re⁃search.A sampler based on BET was designed.PCA and Back-propagation neural network(BPNN)were used in breath pattern recognition.In these simulated experiments,100 healthy breath samples and 100 simulated breath samples were used to validate the breath analysis system.The result of PCA showed that the sensor array showed a distinct response patterns towards two different analytes.With the design of BET sampler,breath samples could be analyzed on-line.The classification accuracy of two kinds of breath samples by pattern recognition is 91.67%.The design of breath analyzer has positive significance for noninvasive detection of diabetes.
作者
张晨生
陆宁
闫梦
武一
花中秋
ZHANG Chensheng;LU Ning;YAN Meng;WU Yi;HUA Zhongqiu(School of Electronics and Information Engineering,Hebei University of Technology,Tianjin 300401,China)
出处
《河北工业大学学报》
CAS
2021年第4期25-30,37,共7页
Journal of Hebei University of Technology
基金
国家自然科学基金(61501167)。
关键词
电子鼻
呼气检测
糖尿病
气体传感器
丙酮
electronic noses
breath detection
diabetes
gas sensor
acetone
