Modern analytical techniques using miniaturized and portable near infrared (NIR) spectroscopy instruments are particularly suited for assessing the authenticity of fishery products since meeting the requirements of rapidity, eco-friendliness, cost-effectiveness, and easiness of application. The objective of the present study was to verify the suitability of use of a portable and ultra-compact NIR spectrometer combined with machine learning to characterize the geographic origin of two octopus species. Replicate NIR spectra (908.1–1676.2 nm) of 118 musky and 29 common octopus specimens (Eledone spp. and Octopus vulgaris) from Portuguese Atlantic or Spanish Mediterranean fishing areas were recorded, pre-processed and elaborated via the following classification algorithms: orthogonal partial least square discriminant analysis (OPLS-DA), logistic regression (LR), random forest (RF), support vector machine (SVM), and multilayer perceptron-artificial neural network (MLP-ANN). When 7-fold cross validation was performed on 75% of data, the results showed that linear tools (OPLS-DA and LR) were the most powerful and stable techniques in recognizing the origin of both octopus species (mean sensitivity, specificity, accuracy, and precision values above 98%). During the external validation phase OPLS-DA, SVM, and MLP-ANN performed better for common octopuses, while LR and MLP-ANN for musky octopuses. The achieved outcomes suggest the combination of portable NIR spectroscopy and machine learning as a promising plan of action to be adopted for the creation of an integrated analytical platform with capabilities for automated data recording, processing, and reporting, which may be helpful for on-site and in-line monitoring of fishery products.
Country of origin label monitoring of musky and common octopuses (Eledone spp. and Octopus vulgaris) by means of a portable near-infrared spectroscopic device / Varra, M. O.; Ghidini, S.; Fabrile, M. P.; Ianieri, A.; Zanardi, E.. - In: FOOD CONTROL. - ISSN 0956-7135. - 138:(2022), p. 109052.109052. [10.1016/j.foodcont.2022.109052]
Country of origin label monitoring of musky and common octopuses (Eledone spp. and Octopus vulgaris) by means of a portable near-infrared spectroscopic device
Varra M. O.;Ghidini S.;Fabrile M. P.;Ianieri A.;Zanardi E.
2022-01-01
Abstract
Modern analytical techniques using miniaturized and portable near infrared (NIR) spectroscopy instruments are particularly suited for assessing the authenticity of fishery products since meeting the requirements of rapidity, eco-friendliness, cost-effectiveness, and easiness of application. The objective of the present study was to verify the suitability of use of a portable and ultra-compact NIR spectrometer combined with machine learning to characterize the geographic origin of two octopus species. Replicate NIR spectra (908.1–1676.2 nm) of 118 musky and 29 common octopus specimens (Eledone spp. and Octopus vulgaris) from Portuguese Atlantic or Spanish Mediterranean fishing areas were recorded, pre-processed and elaborated via the following classification algorithms: orthogonal partial least square discriminant analysis (OPLS-DA), logistic regression (LR), random forest (RF), support vector machine (SVM), and multilayer perceptron-artificial neural network (MLP-ANN). When 7-fold cross validation was performed on 75% of data, the results showed that linear tools (OPLS-DA and LR) were the most powerful and stable techniques in recognizing the origin of both octopus species (mean sensitivity, specificity, accuracy, and precision values above 98%). During the external validation phase OPLS-DA, SVM, and MLP-ANN performed better for common octopuses, while LR and MLP-ANN for musky octopuses. The achieved outcomes suggest the combination of portable NIR spectroscopy and machine learning as a promising plan of action to be adopted for the creation of an integrated analytical platform with capabilities for automated data recording, processing, and reporting, which may be helpful for on-site and in-line monitoring of fishery products.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.