No abstract

a b s t r a c tThe aim of this study was to develop robust chemometric models for the routine determination of dietary constituents of quinoa (Chenopodium quinoa Willd.) using Near-Infrared Transmission (NIT) spectroscopy. Spectra of quinoa grains of 77 cultivars were acquired while dietary constituents were determined by reference methods. Spectra were subjected to multiplicative scatter correction (MSC) or extended multiplicative signal correction (EMSC), and were (or not) treated by Savitzky-Golay (SG) filters. Latent variables were extracted by partial least squares regression (PLSR) or canonical powered partial least squares (CPPLS) algorithms, and the accuracy and predictability of all modelling strategies were compared. Smoothing the spectra improved the accuracy of the models for fat (root mean square error of cross-validation, RMSECV: 0.319e0.327%), ashes (RMSECV: 0.224e0.230%), and particularly for protein (RMSECV: 0.518e0.564%) and carbohydrates (RMSECV: 0.542e0.559%), while enhancing the prediction performance, particularly, for fat (root mean square error of prediction, RMSEP: 0.248e 0.335%) and ashes (RMSEP: 0.137e0.191%). Although the highest predictability was achieved for ashes (SG-

The aim of this study was to develop chemometric models for protein, fat, ashes and carbohydrates contents of quinoa flour using Near-Infrared Transmission (NIT) spectroscopy. Spectra of quinoa flour obtained from grains of 70 different cultivars were scanned while dietary constituents were determined by reference AOAC methods. As a pre-treatment, spectra were subjected to extended multiplicative signal correction (EMSC) with polynomial degree 0, 1 or 2. Next, the Canonical Powered Partial Least Squares (CPPLS) algorithm was applied, and models were compared in terms of accuracy and predictability. For all models, root mean square errors of cross-validation (RMSECV), root meat square errors of prediction (RMSEP) and coefficient of correlation of cross-validation (R CV) were computed. Robust models were obtained when quinoa spectra were pre-processed using EMSC of polynomial degree 2 for both fat (RMSECV: 0.268% and RMSEP: 0.256%) and carbohydrates (RMSECV: 0.641% and RMSEP: 0.643%) following extraction of five CPPLS latent variables. Good coefficients of correlation of prediction (R P : 0.690-0.821) were found for all constituents when models were validated on a test data set consisting of 13 quinoa flour spectra. Thus, good predictions of the dietary constituents of quinoa flour could be achieved by using NIT technology, as implied by the low coefficient of variation of prediction (CV P): 5.64% for protein, 3.88% for fat 7.32% for ashes and 0.80% for carbohydrates contents.

No abstract