Lime juice adulteration detection by spectroscopy and machine learning | ||
| Journal of Food and Bioprocess Engineering | ||
| مقاله 8، دوره 6، شماره 2، بهمن 2023، صفحه 56-62 اصل مقاله (2.27 M) | ||
| نوع مقاله: Original research | ||
| شناسه دیجیتال (DOI): 10.22059/jfabe.2023.364398.1151 | ||
| نویسندگان | ||
| zahra alaei roozbahani* 1؛ Mohsen Labbafi* 2؛ Ali i Aghakhan2؛ Saeed Izadi3 | ||
| 11-Department of Food Science and Technology, College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran 2-Department of Food Science and Technology Standard Research Institute (SRI),Karaj, Iran, z.alaei@standrd.ac.ir | ||
| 21Department of Food Science and Technology, College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran | ||
| 3Simon Fraser University, Canada | ||
| چکیده | ||
| Fruit juices, and especially lime juice, belong to the most targeted food commodities for fraud. Therefore, reliable and cost-effective analytical methodologies need to be developed to guarantee lime juice authenticity and quality. The manifestation of machine learning techniques (MLT) has paved the way for fast and reliable processing and analysis of food and juice data for more effective use of inexpensive, readily available, and easy-to-use equipment such as UV/Vis spectrometers for quality control. The study aimed to investigate UV/Vis spectrometry and MLT to detect at least 10% of water, acid, and sugar added to lime juice. For this purpose, 26 lime samples, including Mexican and Persian lime, were collected from the orchards of four main lime-cultivated areas in Iran to prepare pure lime juice samples (as authentic samples). To investigate adulterated lime juice, four types of treatment were defined by adding acid, sugar, a mix of acid and sugar solution, and water at different volume proportions (10, 20, 30, 40, and 50 % v/v) to pure lime juice samples. Each treatment was repeated eight times. The absorption rate of different adulterated and pure lime juice samples was measured at different wavelengths in the 210–550 nm range. The evaluation results of different MLTs showed that the accuracy of separating samples using absorption data by decision tree (DT), k-nearest neighbor (k-NN), random forest (RF), multilayer perceptron (MLP), and support vector machine (SVM) were 75%, 79%, 80%, 87%, and 92%, respectively. SVM had the highest level of accuracy in separating adulterated lime juice samples. Also, this model’s performance criteria (sensitivity and F-score) were higher than other models for identifying adulterated samples using absorption data. This is the first time that the common adulterations in lime juice were identified by rapid and accessible screening methods using UV/Vis spectroscopy and MLT with high accuracy, precision, and sensitivity. | ||
| کلیدواژهها | ||
| lime juice؛ UV/Vis spectroscopy؛ SVM؛ MLP؛ k-NN | ||
| مراجع | ||
|
| ||
|
آمار تعداد مشاهده مقاله: 409 تعداد دریافت فایل اصل مقاله: 492 |
||