An electronic nose system supported by machine learning techniques for rapid detection of aspergillus flavus in pistachio

Pistachio is market appealing due to its high nutritional value, low calorie and fat content, high anti-oxidant properties, and special aroma and flavor. However, this fruit is prone to various pathogenic factors, including fungi. This may lead to the production of highly toxic Aflatoxins. In this respect, the detection of fungal pathogens infection in the early stages of pistachio production is a major challenge in food security to mitigate losses as much as possible. In this study, an electronic nose (E-nose) was employed as a non-destructive and fast method for early detection of fungal infection on pistachio by Aspergillus Flavus fungus synthetically. To prepare experimental treatments, three treatments of 10², 10⁴, and 10⁶ spores were considered. An electronic nose system consisting of eight metal semiconductor sensors was developed to assess the odor of the samples. Three methods of principal component analysis (PCA), linear discriminant analysis (LDA), and support vector machine (SVM) were applied for classifications. The results showed that the electronic nose system was able to separate different concentrations of contamination from each other with an accuracy of 100%. In addition, the obtained results represented that infected pistachio samples can be successfully differentiated starting from the third day of the infection period. Notably, LDA and SVM performed better than the PCA method with 100% accuracy. Therefore, the electronic nose is a practical and beneficial tool for diagnosing the pistachio aflatoxin fungal infection. Therefore, the electronic nose can be a practical and beneficial tool that replaces traditional methods for diagnosing pistachio aflatoxin fungal infection.