Data fusion approaches in spectroscopic characterization and classification of PDO wine vinegars

Abstract

Spain is one of the major producers of high-quality wine vinegars having three protected designations of origin (a.k.a. PDOs): "Vinagre de Jerez", "Vinagre de Condado de Huelva" and "Vinagre de Montilla-Moriles". Their high prices due to their high quality and their high production costs explain the need for developing an adequate quality control technique and the interest in extensive characterization in order to capture the identity of each denomination. In this framework, methodologies based on non-targeted techniques, such as spectroscopies, are becoming popular in food authentication. Thus, for improving vinegar quality assessment, fusion of data blocks obtained from the same samples but different analytical techniques could be a good strategy, since the quantity and quality of sample knowledge could be enhanced providing new insights into the differentiation of vinegars. Therefore, the aim of this manuscript is the development of a multi-platform methodology and a model able to classify the Spanish wine vinegar PDOs. Sixty-five PDO wine vinegars were analyzed by four spectroscopic techniques: Fourier-transform mid-infrared spectroscopy (MIR), near infrared spectroscopy (NIR), multidimensional fluorescence spectroscopy (EEM) and proton nuclear magnetic resonance (1 H-NMR). Two different data fusion strategies were evaluated: Mid-level data fusion with different preprocessing, and Common Component and Specific Weights analysis multiblock method. Exploratory and classification analysis on the data from individual techniques were also performed and compared with data fusion models. The data fusion models improved the classification, providing a more efficient differentiation, than the models based on single methods, and supporting the approach to combine these methods to achieve synergies for an optimized PDO differentiation.

Discrimination of Gentiana and Its Related Species Using IR Spectroscopy Combined with Feature Selection and Stacked Generalization

“…In recent years, SVM has successfully been applied to a number of applications, such as classification of species or geographical origin traceability of food [53,68,69]. It is important to note that SVM can achieve high classification accuracy whlie using a small number of training samples [56,67]. Additionally, it is also a suitable classifier for high-dimensional data [53,69].…”

“…As we know, the data fusion (low-level and mid-level) approaches present a fusion of all variables or most important variables (feature variables) to create a model in order to exploit the synergy of the multispectral information to obtain an optimized model [53][54][55][56]. However, the calculation time might be higher when increasing variables.…”

Nontargeted Analytical Methods as a Powerful Tool for the Authentication of Spices and Herbs: A Review

“…Therefore, the differences are not necessarily based on environmental factors and agricultural production, but on the genetic variations inherent in the species. For this, data fusion using the composition of metabolites and NIR information can generate more accurate answers, as was reported in the wine PDO identification (Ríos-Reina, Callejón, Savorani, Amigo, & Cocchi, 2019). Regarding spices, only one work was found, which reports the use of the fusion of NMR and NIR data to identify spices adulterated with Sudan III and IV (Di Anibal, .…”

Use of Spectroscopic Techniques for a Rapid and Non-Destructive Monitoring of Thermal Treatments and Storage Time of Sous-Vide Cooked Cod Fillets

“…The rest of the fish (96 samples) were divided into 8 groups, each group of 12 samples. The samples were cooked at different temperatures (50,60,70 or 80 • C) for 5 or 10 min. Before cooking, the samples were vacuum-packed and then cooked in a water bath preheated at the desired temperature.…”