Application of support vector machines to 1H NMR data of fish oils: methodology for the confirmation of wild and farmed salmon and their origins

Abstract

Support vector machines (SVMs) were used as a novel learning machine in the authentication of the origin of salmon. SVMs have the advantage of relying on a well-developed theory and have already proved to be successful in a number of practical applications. This paper provides a new and effective method for the discrimination between wild and farm salmon and eliminates the possibility of fraud through misrepresentation of the country of origin of salmon. The method requires a very simple sample preparation of the fish oils extracted from the white muscle of salmon samples. (1)H NMR spectroscopic analysis provides data that is very informative for analysing the fatty acid constituents of the fish oils. The SVM has been able to distinguish correctly between the wild and farmed salmon; however ca. 5% of the country of origins were misclassified.

Quality of farmed and wild sea bass lipids studied by 1H NMR: Usefulness of this technique for differentiation on a qualitative and a quantitative basis

“…To do so, the development of reliable methods is necessary. Many recent papers on this topic involve the use of sophisticated statistical approaches (Axelson, Standal, Martinez, & Aursand, 2009;Masoum et al, 2007;Rezzi et al, 2007); however, some authors invoke the need to develop reliable chemical and instrumental methods without the need for complicated statistical approaches (Frankel, 2010).…”

Identification of the farm origin of salmon by fatty acid and HR 13C NMR profiling

“…Optimal application of these analytical techniques, in particular of the complex information contained in the HR NMR spectra, requires the use of advanced chemometrics in order to obtain correct classifications, such as principal component analysis (PCA), probabilistic neural networks (PNN) (Aursand, Standal, & Axelson, 2007;Specht, 1990), Bayesian belief networks (BBN) (Glover et al, 2008;Heckerman, Geiger, & Chickering, 1995) or support vector machines (SVM) (Masoum et al, 2007).…”

An Emerging Powerful Technique: NMR Applications on Quality Assessments of Fish and Related Products

“…Fish feed and raw ingredients [3] HF-NMR Quantification of betaine in mussels Mussels [28] Quantification of n-3 fatty acids, anserine, and lactate Salmon and fish oils [48,59] Study of changes in bioactive components Cod [29] Low molecular weight metabolites Cod and haddock [31] Post-mortem metabolic changes in muscle by 31 Fish oils [34,35] Omega-3 fatty acid content by HR-MAS Salmon [62] Metabolic condition changes upon thermal stress Steelhead trout [37] Metabolic response to long-term handling Salmon [38] Metabolic differences between healthy and hepatic tumour tissues Flatfish [39] Metabolic markers upon toxic exposure Minnow, Chinook salmon, medeka, and rainbow trout [40][41][42][43] Authenticity Wild-farmed salmon fish oils [2,47] MRI Backbone deformations Salmon [5] Belly bursting Herring [52] Salt content by 1 H and 23 Na MRI Salmon and cod [57] Abbreviations: LF-NMR, low-field nuclear magnetic resonance; HF-NMR, high-field NMR; MRI, magnetic resonance imaging; HR-MAS, high resolution-magic angle spinning.…”

“…However, the present NMR methods for fish authentication should still be improved, since it is not yet an internationally accepted, validated, and consistent method that emerged from the attempts made so far [46]. These attempts, among many others, include the study of 1 H NMR data evaluated by using advanced data mining methods that allowed a successful identification of wild and farmed salmon [2]. For fish oil and related health products, Aursand et al [47] showed that 13 C NMR data, as examined by multivariate analysis techniques, could provide information about the nature, composition, refinement, and adulteration of the products.…”