阿韦纳
偏最小二乘回归
近红外光谱
光谱学
红外光谱学
分析化学(期刊)
近红外反射光谱
化学
食品科学
葡聚糖
均方根
数学
植物
色谱法
生物
统计
生物化学
物理
光学
量子力学
有机化学
作者
Gracia Montilla‐Bascón,Paul R. Armstrong,Rongkui Han,Mark E. Sorrells
标识
DOI:10.1177/0967033517709615
摘要
Whole oat has been described as an important healthy food for humans due to its beneficial nutritional components. The positive health benefits of consuming oats as a whole-grain food are attributed in part to β-glucan, which has outstanding functional and nutritional properties. Near infrared reflectance spectroscopy is a powerful, fast, accurate and non-destructive analytical tool that can be substituted for some traditional chemical analysis. A total of 1728 single intact groats of six different oat varieties were scanned by near infrared spectroscopy to develop non-destructive predictions for (1,3;1,4)-β-D-glucan (β-glucan), protein and oil content in groats. Prediction models for single kernels were developed using partial least squares regression. Regression parameters between the chemical values, determined by wet-lab reference methods, and the predicted values determined from near infrared spectra, were verified by cross-validation and against data from a set of independent samples. The cross-validation correlation coefficients ( R 2 CV ) for β-glucan, protein and oil were 0.83, 0.72 and 0.92, respectively, the root-mean-square error ranged from 0.25% to 0.60% for all compounds. Independent validation data had r 2 values ranging from 0.69 to 0.95; root-mean-square error of prediction values (RMSEP) values were equal to or less than 0.52%, 0.62% and 0.27% for β-glucan, protein and oil, respectively. The data indicated that non-destructive screening of β-glucan, protein and oil contents in single kernels of dehulled oat grains from their near infrared spectra could be successfully used in breeding programs.
科研通智能强力驱动
Strongly Powered by AbleSci AI