Rice varietal integrity and adulteration fraud detection by chemometrical analysis of pocket-sized NIR spectra data

Abstract

Rice consumption is on the increase in Ghana and Africa as a whole. This has resulted in mislabelling and adulteration fraud that is affecting many players in the rice value chain. This research attempts to provide a user-friendly and reliable onsite analytical tool using a pocket-sized NIR spectrometer and multivariate data for detecting rice integrity and fraud. A total of 112 rice samples were made up of three different categories; 36 samples of the Jasmine variety, 36 samples of the Agra variety, and 40 adulterated Jasmine with Agra (10–40% w/w) were used. Multivariate spectral data analysis was used to model the best technique for simultaneous identification and quantification of rice variety integrity and fraud. For the optimum identification of the challenge, rice samples powdered had a better performance compared to rice grain, at an accuracy of 98% in both calibration and prediction sets after modelling with the SD-PLSDA mathematical algorithm. For the quantification of adulteration fraud, the amount of substitution challenge, higher accuracy was found when powdered rice samples were used compared to rice grain. The best result was obtained by Si-PLS at R2 = 0.94 in both calibration and prediction sets, with RMSEP and RMSECV at 0.13 and 0.16 respectively. This study has shown that pocket-sized NIR spectroscopy could provide a promising tool for easy, rapid, and onsite early detection of rice integrity and varietal adulteration fraud in the rice value chain, and it has the potential to be incorporated into smartphone devices. This would be very useful to breeders, rice farmers, millers, and consumers alike in developing countries where laboratory infrastructure is a major challenge.

Keywords

• Rice
• Fraud
• Identification
• Quantification
• Multivariate analysis