Abstract Adulteration of raw milk with whey is difficult to evaluate because both have great physical and chemical similarities. Herein, we proposed a rapid and cost-effective method to detect and quantify the presence of whey in milk using Raman spectroscopy, while dispensing any step of sample preparation. Principal Component Analysis (PCA) was performed for outlier detection and Partial Least Squares regression (PLS) for prediction of whey content. Since single vibrational modes have been not identified in order to discriminate the different levels of whey addition, we proposed a calibration model from the full Raman spectra using PLS regression. We explored the idea that small differences in spectroscopic data could be useful in this evaluation by using multivariate chemometric methods. As result, the correlation coefficient between the Raman method and the experimental values was R² = 0.988 for the validation samples. A paired t-test revealed a p-value greater than 0.5, pointing out that no significant statistical differences between the proposed Raman method and the reference values. The approach proposed in this study shows to be quite useful for the evaluation of milk adulteration by whey addition, thus contributing to quality control of milk in dairy industry.

中文翻译：

---

拉曼光谱作为原料奶中乳清定量的快速工具

摘要 原料奶中的乳清掺假很难评估，因为两者在物理和化学上有很大的相似性。在此，我们提出了一种快速且经济高效的方法，可使用拉曼光谱检测和量化牛奶中乳清的存在，同时免除样品制备的任何步骤。主成分分析 (PCA) 用于异常值检测和偏最小二乘回归 (PLS) 用于预测乳清含量。由于尚未确定单一振动模式以区分不同水平的乳清添加，我们使用 PLS 回归从完整的拉曼光谱中提出了一个校准模型。我们通过使用多元化学计量学方法探索了光谱数据的微小差异在此评估中可能有用的想法。结果，对于验证样品，拉曼方法与实验值之间的相关系数为 R² = 0.988。配对 t 检验显示 p 值大于 0.5，表明提出的拉曼方法与参考值之间没有显着的统计差异。本研究中提出的方法对于评估添加乳清造成的牛奶掺假非常有用，从而有助于乳制品行业的牛奶质量控制。