Non-Invasive Breath Analysis for Lung Cancer Screening Using CNN

Current diagnostic techniques tend to be invasive, time-consuming, expensive, and may not consistently yield accurate results. To enable early detection, recent studies have concentrated on noninvasive methods, such as identifying volatile organic compounds (VOCs) in human breath. Variations in the types and levels of VOCs in exhaled breath can signal the presence of lung cancer, making this approach advantageous compared to conventional diagnostic procedures. The system operates in three phases: breath collection, VOC detection, and data analysis[1]. When the user exhales into the mouthpiece, a flow sensor detects the breath and signals the Arduino Nano, which controls solenoid valves to direct the sample into a storage chamber [2]. Simultaneously, the sample was stored in a Tedlar bag for further analysis. The collected VOC data from database were analyzed using a convolutional neural network (CNN) algorithm, which classifies lung disease stages based on VOC patterns, providing a user- friendly diagnostic output.