Abstract Secondary metabolites are important components in terms of nutrition and health. Carotenoids and tocopherols, two groups of the fat-soluble components, are also included in this category. There is an increasing interest in the detection of secondary metabolites with near-infrared spectroscopy. However, the number of scientific studies for the detection of these components, especially for tocopherols in corn flour or oil samples by near-infrared reflectance spectroscopy is rather limited. This study was carried out to determine the amount of carotenoids and tocopherols in flour and oil samples of 250 different maize genotypes by near-infrared reflectance spectroscopy using the partial least squares regression modeling method. Liquid chromatography mass spectrophotometry was used as a reference method in order to determine the contents of five carotenoids and four tocopherol subcomponents. The estimation models were created by using the spectral data collected from ground samples, and oil samples extracted from the same flour; along with the results of the reference analysis. The reliability of these models was tested by external validation (n = 50). The prediction models generated by the spectra taken from corn flour yielded more successful results than the models created with the spectra taken from the oil samples. Among the models compared, the one developed with the spectra taken from flour samples for lutein was the most successful. It is seen that the estimation models generated from flour samples can be used for screening purposes, though different approaches are needed to increase the success of models.

中文翻译：

---

使用近红外光谱法测定玉米粉和油样品中的类胡萝卜素和生育酚含量

摘要 次级代谢产物是营养和健康的重要组成部分。类胡萝卜素和生育酚这两组脂溶性成分也包括在这一类别中。人们对使用近红外光谱检测次级代谢物越来越感兴趣。然而，通过近红外反射光谱检测这些成分，特别是玉米粉或油样品中的生育酚的科学研究数量相当有限。本研究旨在通过近红外反射光谱法使用偏最小二乘回归建模方法测定 250 种不同玉米基因型的面粉和油样品中类胡萝卜素和生育酚的含量。液相色谱质谱法被用作参考方法，以确定五种类胡萝卜素和四种生育酚亚成分的含量。估计模型是通过使用从地面样品收集的光谱数据和从同一面粉中提取的油样品创建的；以及参考分析的结果。这些模型的可靠性通过外部验证（n = 50）进行了测试。由玉米粉光谱生成的预测模型比使用油样光谱创建的模型产生了更成功的结果。在比较的模型中，使用从面粉样品中提取的叶黄素光谱开发的模型是最成功的。可以看出，从面粉样品生成的估计模型可用于筛选目的，