Research Gas Recognition Algorithm Based On Lung Algorithm Gas Detection System

In recent years, lung cancer is a common cancer diseases threaten the lives and health of people. Along with the worsening of the domestic air, dust storms, smog has increased, lung cancer incidence has increased significantly. If Lung cancer was diagnosed in early stage and the surgical removal of the positive treatment can improve the survival rate of patients, at present the study is the lack of non-invasive, low-cost, rapid lung cancer testing equipment. Breathing gas lung cancer detection based on porphyrin sensor system can respond to the patient’s breath gas and sensor array response difference map implementation of noninvasive, rapid screening for lung cancer. Difference map is the key to determining the difference between different types of lung cancer and iconic gas concentration. In order to design a signature based on difference maps implement put forward recognition algorithm of different lung cancer gas quantitative and qualitative pattern has lung cancer early clinical screening potential application prospects.This paper presents a respiratory gas used in embedded systems qualitative detection of lung cancer recognition algorithm. The algorithm combines fuzzy evaluation criteria template matching recognition of different lung cancer iconic gas weighted template matching to determine the type of gas to be measured. Then Qt platform programming Linux system, transplanted to the embedded load detecting device applications. After the qualitative identification, extraction of color properties characteristic component on sensitive points, using different identification methods for quantitative gas concentration. Specific research work includes the following aspects:(1) Under the theory and fuzzy template matching, using a sensor array dynamic response curve porphyrin unit to determine the different lung iconic gas in response to the number and position sensitive, to generate a template map and then be tested by Atlas map template point weighted template matching, by a similar measure, determine the type of gas to be measured.(2) Different lung cancer iconic gas and lung cancer patients exhaled instances to identify and recognize the traditional cluster analysis results were compared, the results show that the qualitative identification algorithm design has better recognition performance, low detection limit, and to achieve the distinction between lung cancer patients and healthy volunteers exhaled breath gas.(3) Based on embedded lung respiratory gas detection system built with Linux operating system to Qt-Creator integrated development platform for the qualitative identification procedures, the algorithm will be loaded into the image processing software part of the realization of lung cancer iconic gas species identification and display.(4) After the qualitative identification, extraction of gas in response to the iconic color properties characteristic component sensitive points on the gas concentration quantitative identification. Using discriminant analysis, support vector machine, BP neural network hue and saturation characteristics of the volume analysis, support vector machine recognition showed good recognition performance, and the algorithm is better than discriminant analysis. BP neural network analysis results showed that within the extracted feature amount of concentration to identify errors in the permissible range.