Determination of fungus infected apples by using electronic nose

Electronic noses mimic the human olfactory system using machine learning algorithms to enable odor identification and classification. Equipped with sensors, they can identify gases imperceptible to the human nose, allowing for early detection of issues such as fruit decay. Given the apple's susceptibility to fungal deterioration during storage, the electronic nose provides an advantage in early diagnosis of these problems. In this study, a dataset was created with a total of 160 samples, including an equal number of fresh apples and apples injected with Aspergillus niger, Penicillium expansum, and Penicillium crutosum fungi. To determine apple characteristics, specific data were eliminated, and sequential processes of feature extraction and linear discriminant analysis were applied. Training and test accuracies were determined using K-nearest neighbors, decision trees, support vector machines, and naive Bayes classification methods. In the 4-class classification of fresh apples and apples injected with Aspergillus niger, Penicillium expansum, and Penicillium crutosum fungi, the K-nearest neighbors classification method achieved the highest test accuracy of 99.33%. The study results demonstrate that the electronic nose enables early detection of decay in apples.