Optimization of Metal Oxide Nanosensors and Development of a Feature Extraction Algorithm to Analyze VOC Profiles in Exhaled Breath

Exhaled volatile organic compounds (VOCs) have been identified as biomarkers for different diseases. Electronic noses (e-Noses) utilizing metal oxide (MOX) sensors for VOC detection are sensitive to a range of gases and offer rapid detection and portability. E-Noses have integrated feature extraction algorithms, but in-house systems do not, and manual extraction is time-consuming and prone to error. MOX sensor arrays have been previously tested using synthetic VOCs but there are limited studies seeking to optimize exhaled breath analysis. The goal of this study is to develop an automated feature extraction algorithm to optimize SnO
₂
nanosensor parameters and breath sampling methods. Python was used to develop an algorithm that can extract peak–peak value, relative abundance, slope, and other sensor features. After verifying algorithm performance, sensor operating parameters including heater/sensor voltages were optimized. Optimal parameters were utilized to analyze simulated breath with varying humidity levels. Exhaled breath sampling protocols were explored by testing different sensor housing designs, fractionating breath, and standardizing collection by volume. Optimal parameters for SnO
₂
include a heater voltage equal to 2 V and a sensor voltage of 0.8 V, and the sensor could distinguish VOC profiles in simulated breath independent of varying humidity levels. Sensor testing with real breath samples showed no increase in reproducibility when fractionating breath, and that sampling 24 L provided the highest sensitivity. The SnO
₂
sensors were utilized to analyze breath samples from three volunteers, and the results showed high intrasubject reproducibility as well as separation between subjects.