Original Research ArticleRecognition of organic rice samples based on trace elements and support vector machines
Introduction
Health concerns and environmental considerations related to the use of pesticides, hormones and veterinary drugs in conventional farming practices have raised public attention to the production and consumption of organic products. Organic farming is an environmentally sustainable approach with viable solutions to a range of problems associated with conventional food production, rural development and animal welfare. In the specific case of organic cereal production, current legislation restricts the use of synthetic fertilizers, pesticides and insecticides. Farming practices should be based on the use of compost or organic waste, readily soluble mineral fertilizers and biological pest control (Capuano et al., 2012).
Due to lower production yields and higher certification costs, organic products tend to retail at higher prices than their conventional counterparts. This fact, associated with their increasing popularity, makes organic products susceptible to fraud. One possible way to differentiate among organic and conventional products is to monitor their chemical composition. Under this perspective, the authentication of organic foods becomes an analytical problem. The studies presented here deal with the authentication of organic rice.
Several analytical strategies exist for the authentication of organic foods of plant origin. These include the measurement of stable isotope ratios such as 15N/14N (Kelly et al., 2005, Rossmann, 2001), metabolomics (Zorb et al., 2006), analysis of phenolic compounds (Wang et al., 2008), copper chloride crystallization (Huber et al., 2010) and infrared spectrometry (Cozzolino et al., 2009). The approach presented here is based on quadrupole-inductively coupled plasma-mass spectrometry (q-ICP-MS) and the differences found in the elemental composition of organic and conventional rice. Although ICP-MS has been applied to the determination of metals in rice samples (Capuano et al., 2012, Cheajesadagul et al., 2013, Shen et al., 2013), our literature search reveals no previous reports for the purpose at hand. Our work considers the spectral profiles of rice samples resulting from the determination of three major (g kg−1) and seventeen minor (mg kg−1 – mg kg−1) elements. Data analysis was carried out with the aid of a data mining technique known as support vector machine (SVM). Supervised and unsupervised chemometric tools have been used for food authenticity and traceability (Barbosa et al., 2014a, Barbosa et al., 2014b, Batista et al., 2012, Niu et al., 2011). SVM is a supervised learning model with associated learning algorithms that analyze data and recognize patterns often used for classification and regression analysis (Tan et al., 2006). A literature search has not revealed any reports of its application to data processing of chemical profiles from rice samples. Herein, we demonstrate that the combination of q-ICP-MS with SVM provides a robust analytical tool for the authentication of organic rice samples.
Section snippets
Rice sampling
Certified organic (n = 17) and conventional (n = 33) rice samples were purchased from different Brazilian producers (50 different brands) in several cities in the states of Rio Grande do Sul, Santa Catarina, Minas Gerais, Goiás and Tocantins. All organic rice samples were certified by the Brazilian IBD-Agricultural and Food Inspections and Certifications that is accredited by the International Federation of Organic Agriculture Movements. Rice samples (15 g) were separated by quartering as described
Analytical figures of merit
Nineteen elements were determined in rice samples. These included toxic elements (As, Pb and Cd), essential elements (Cu, Zn, Mg, P, Mo, Mn, Se, Co, Cr, Fe, and Ca) and other elements such as B, Ba, Rb, La and Ce. Their choice was based on previous reports showing considerable concentrations in both types of samples, i.e. conventional and organic rice (Batista et al., 2010). q-ICP-MS was selected for the analysis due to its well-known sensitivity toward the targeted elements. Table 2 summarizes
Conclusions
To our knowledge, this manuscript is the first to demonstrate that the use of q-ICP-MS associated with SVM is a highly effective method for classifying organic rice samples. With the proposed procedure, it was possible to predict the authenticity of organic rice samples with an accuracy of 98% when using the 19 original elements. An accuracy of 96% was found when only the levels of elements Ca and Cd was selected.
Conflict of interest
None.
Acknowledgements
Authors are grateful to Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for financial support and fellowships.
References (27)
Support vector machines combined with feature selection for breast cancer diagnosis
Expert Syst. Appl.
(2009)- et al.
Recognition of the geographical origin of beer based on support vector machines applied to chemical descriptors
Food Control
(2012) - et al.
The use of advanced chemometric techniques and trace element levels for controlling the authenticity of organic coffee
Food Res. Int.
(2014) - et al.
Speciation of arsenic in rice and estimation of daily intake of different arsenic species by Brazilians through rice consumption
J. Hazard. Mater.
(2011) - et al.
Multi-element determination in Brazilian honey samples by inductively coupled plasma mass spectrometry and estimation of geographic origin with data mining techniques
Food Res. Int.
(2012) - et al.
Mid infrared spectroscopy and multivariate analysis: a tool to discriminate between organic and non-organic wines grown in Australia
Food Chem.
(2009) - et al.
Tracing the geographical origin of food: the application of multi-element and multi-isotope analysis
Trends Food Sci. Technol.
(2005) - et al.
The use of inductively coupled plasma mass spectrometry (ICP-MS) for the determination of toxic and essential elements in different types of food samples
Food Chem.
(2009) - et al.
Inductively coupled plasma-mass spectrometry (ICP-MS) and -optical emission spectroscopy (ICP–OES) for determination of essential minerals in closed acid digestates of peanuts (Arachis hypogaea L.)
Food Chem.
(2012) - et al.
Trace elements determination in bovine semen samples by ICP-MS and data mining techniques for identification of bovine class
J. Dairy Sci.
(2012)
The use of decision trees and Naïve Bayes Algorithms and trace element patterns for controlling the authenticity of free-range-pastured hens eggs
J. Food Sci.
Survey of 13 trace elements of toxic and nutritional significance in rice from Brazil and exposure assessment
Food Additiv. Contaminants: Part B
Cited by (68)
Proven traceability strategies using chemometrics for organic food authenticity
2024, Trends in Food Science and TechnologyAuthentication of beef cuts by multielement and machine learning approaches
2023, Journal of Trace Elements in Medicine and BiologyHoney authentication and adulteration detection using emission – excitation spectra combined with chemometrics
2023, Spectrochimica Acta - Part A: Molecular and Biomolecular SpectroscopyA comparative study regarding the adulteration detection of honey: Physicochemical parameters vs. impedimetric data
2023, Current Research in Food ScienceRice authentication: An overview of different analytical techniques combined with multivariate analysis
2022, Journal of Food Composition and AnalysisCitation Excerpt :Statistical techniques including PCA, SIMCA, HCA, and KNN of elemental fingerprints showed great potential to discriminate organic from ordinary rice samples (Borges et al., 2015). In a similar approach, 19 elements including As, B, Ba, Ca, Cd, Ce, Cr, Co, Cu, Fe, La, Mg, Mn, Mo, P, Pb, Rb, Se and Zn were analyzed in conventional (n = 33) and organic (n = 17) rice samples by quadrupole ICP-MS. Ca and Cd concentration exhibited the maximum variation between organic and conventional grown samples and the authenticity of organic samples with an accuracy of 96% was achieved using only these two elements (Barbosa et al., 2016). In a different approach, the feasibility of stable carbon and nitrogen isotopes was examined for the authentication of organic rice procured from the local Korean market.
Determination of total protein and wet gluten in wheat flour by Fourier transform infrared photoacoustic spectroscopy with multivariate analysis
2022, Journal of Food Composition and AnalysisCitation Excerpt :PLSR decomposes dependent variable X and independent variable Y matrices into principal components at the same time and extracts comprehensive variables with strong explanatory ability of dependent variables to maximize the correlation between principal components and concentrations, thus overcoming the negative impact of multiple correlations of variables and further improving the reliability of the model. SVM is based on the strategy of structural risk minimization (SRM) and has excellent training and generalization abilities (Barbosa et al., 2016). Input vectors are mapped to a high dimensional space where a maximal separating hyperplane is constructed.