Discriminability Assessment of MOS Sensors to Develop a Portable Electronic Nose System for Non-Destructive Quality Assessment of Medicinal Plants

In this study, the discriminatory power of Metal Oxide Semiconductor (MOS) gas sensors was evaluated to develop a portable electronic nose (e-nose) system for clustering different types of saffron samples based on their Volatile Organic Components (VOCs). The system was comprised of ten MOS gas sensors, direct headspace sampling, microcontroller devices, and a laptop computer coupled with multivariate computational tools. Eleven saffron samples were procured from different geographical origins for the experiments. Principal Component Analysis (PCA) and Hiricultural Cluster Analysis (HCA) models were applied for sample clustering and for demonstrating the discriminatory power of the gas sensors as well. The quality assessment of the samples was also performed by the standard laboratory method (ISO3632). The gas sensors data were acquired at 350-360 seconds after the samples were exposed to the sensors. Results of the PCA and HCA analysis of the sensors data indicated that the saffron samples were divided into three main clusters. Also, it was found that the discrimination power of the sensors was different and the possibility of removing sensors with low discriminatory power (2 sensors) was provided. Results of laboratory analysis (destructive method) were obtained in accordance with the classification results of the e-nose data analysis. Using the proposed method to find the most effective MOS sensors and eliminating the redundant sensors can help to reduce the development costs of the electronic nose systems and the processor input data. It also increases the classification accuracy of the e-nose system in the quality control of medicinal plants.