Evolving Window Zone Selection method followed by Independent Component Analysis as useful chemometric tools to discriminate between grapefruit juice, orange juice and blends

This study investigates the use of high resolution H-1 NMR as a suitable alternative to the standard chromatographic method for the determination of adulteration of orange juice (Citrus sinensis) with grapefruit juice (Citrits paradisi) based on flavonoid glycoside content. Fifty-nine orange juices (OJ), 23 grapefruit juices (GJ) and 10 blends (OG), obtained from local retail outlets were used to assess the performance of the H-1 NMR method. The work presented here introduces the Evolving Window Zone Selection (EWZS) function that holds promise for the automatic detection of spectral regions tailored to discriminate predefined groups. This technique was applied on the pre-processed H-1 NMR spectra of the 92 juices. Independent Component Analysis (ICA) is a good alternative to Principal Component Analysis (PCA) for recovering linearly-mixed unobserved multidimensional independent signals and has been used in this study to build supervised models that classify the samples into three categories, OJ, GJ, OG. The regions containing the known flavonoid glycoside markers were selected as well as another zone containing the signals of sucrose, a-glucose and other components that were tentatively attributed. ICA was applied on three different groups of selected variables and showed good results for both discrimination and interpretation of the signals. Up to 97.8% of the juices were correctly attributed. This method gave better results than the commonly used PCA method. In addition, the time required to carry out the H-1 NMR analysis was less than half the time of the standard chromatographic method. (C) 2007 Elsevier B.V. All rights reserved.