Abstract
Increasing demand for plant-based proteins has elicited suspicions of fraudulent practices among some producers of melon seed powder. The powder is suspected to be adulterated with carbohydrate-rich foods including maize, dried-milled cassava (gari) and soy flour to boost its quantity for increased profit. This study aimed to develop robust models using near-infrared spectroscopy (NIRS) for rapid authentication of melon seed powder. Melon seed powder samples containing 0–50 %w/w maize, gari and soy were prepared in triplicates and scanned using a handheld NIRS device. Proximate composition and color were significantly (p<0.05) affected by adulterant presence in melon seed powder. Classification models using linear discriminant analysis yielded accuracies of adulterant classification that were further confirmed by the high sensitivity, specificity and precision of the models (higher than 80 % in all cases). Low errors (RMSEC 3.28, RMSECV 3.92, RMSEP 6.87) % w/w, limits of detection and quantification and high coefficients (R2 0.96, R2CV 0.94, R2PRED 0.81) % w/w, indicative of reliable partial least squares models were obtained for the prediction of adulterants, protein, carbohydrate, total colour change, fat and lightness. Overall, Savitisky-Golay smoothing filter yield the best results in all LDA analysis. NIRS is a viable option for rapid non-invasive authentication of melon seed powder. The handheld device presents an added advantage of remote analysis. Instruments with other wavelength ranges can be explored in a similar approach study but with a higher sample size for more rigorous applications.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have