Lung Cancer Detection Utilizing Mixed Sensor Based Electronic Nose

Abstract

The lungs are the most important organ of the respiratory system. The volatile organic compounds found in breath, which usually originate from the blood and enable us to observe different processes in the body, are carried to the lungs through the blood and then exhaled by breath. In this study, a mixed sensor based electronic nose circuit was developed using eight metal oxide semiconductors and 14 Quartz Crystal Microbalance gas sensors. A total of 100 volunteers participated in the study, including 20 healthy volunteers who did not smoke, 20 healthy volunteers who smoked, and 60 lung cancer volunteers. Throughout this study, 338 experiments were conducted using breath samples. Data dimension reduction was achieved using linear discriminant analysis and principal component analysis algorithms. The individual classification accuracies for the metal oxide semiconductor and quartz crystal microbalance sensor data are 81.54% and 73.18%, respectively. Upon combining the sensor data, a noticeable increase in accuracy of 85.26% was observed. In this study, the performance of the developed system was enhanced using principal component and linear discriminant analyses. While the highest classification accuracy increased to 88.56% with the feature matrix obtained using the principal component analysis method, this value was obtained with 94.58% accuracy using the linear discriminant analysis method.