Study On Design And Algorithm Of Electronic Nose For Lung Cancer Detection

This thesis designed and construct an electronic nose sensor array for lung cancer identification to achieve non-invasive detection of lung cancer.Conventional electronic nose technology and pre-concentration technology were combined to form an electronic nose system.In order to improve the response of sensor,the optimal sweeping(sampling)speed of the electronic nose system was found using Newton method,and patients with lung cancer and those with non-lung cancer(including healthy people and patients with other lung diseases)were classified and identified.The main contents of the thesis are as follows:(1)Aiming at the common volatile metabolites in the exhaled gases of lung cancer patients,this thesis selected various types of sensors to form the sensor array.In addition,in order to improve the anti-interference ability of sensor array,temperature,humidity,air pressure,working voltage and heating voltage were also added.A total of25 sensors are used in sensor array(a total of 29 signals).In order to reduce the volume of sensor chamber,a set of sensor chamber conforming to the sensors was designed using mechanical process,and Teflon is coated inside the sensor chamber to reduce odor interference.In addition,a corresponding pre-concentration system was designed to improve the detection capability of the electronic nose system.(2)The connectivity and diffusion of conformal sensor chamber were analyzed by FLUENT simulation software.On the basis of lung cancer detection electronic nose system and sensor chamber,gas chromatography-mass spectrometry(GC-MS)was used to test the performance of adsorption tube at different sweeping speeds.According to the change of sensor comprehensive response,Newton method was used to determine the optimal sweeping speed of sensor chamber and lung cancer detection electronic nose system.Also,the exhaled gases response of human verified the optimal speed.(3)Based on the designed alveolar gas collector which is suitable for collecting human exhaled breath and lung cancer detection electronic nose,experiments and comparisons were carried out on healthy people,patients with lung cancer and patients with diseases of non-lung cancer.The classification and regression tree(CART),random forest(RF)and extreme gradient boosting(XGBoost)algorithm were used to classify and identify the experimental samples.The accuracy of identification has reached 81.6%.The classification results of lung cancer patients by electronic nose and enrichment-desorption unit(EDU)combined with PEN3 system developed by the German Airsense company were analyzed,respectively.And it is proved that the lung cancer identification performance of the designed lung cancer detection electronic nose is better than that of EDU + PEN3 system.The innovation of the thesis:(1)The concentration of volatile organic compounds in the exhaled breath of patients with lung cancer is low,so a pre-concentration system was introduced into the conventional electronic nose to improve the identification capability of the electronic nose system for lung cancer detection.(2)Based on the designed lung cancer detection electronic nose system,the sensor comprehensive response of the conformal sensor chamber was used to to find the best sweeping speed of the system to further improve the discrimination effect of the system.