近红外高光谱成像技术在筛查保健食品中违禁添加抗炎药物双氯芬酸钠的研究

双氯芬酸钠是一种非甾体抗炎药，抗炎症效果较好，抗炎化学药物双氯芬酸钠有可能违禁添加到缓解炎症保健食品中。目前已有一些文献研究水、肉及牛奶等中的药物双氯芬酸钠，主要利用高效液相色谱、表面增强拉曼及电化学等方法，需要相对复杂的预处理过程，操作步骤相对复杂，耗时。为了快速、无损检测保健食品中违禁添加物的成分，探索建立采用1 000～2 524 nm波段或特征最优波段的光谱作为自变量，将近红外高光谱成像技术结合化学计量学分析的方法定量分析缓解炎症保健食品中可能添加的不同浓度的双氯芬酸钠。研究主要运用了8种光谱预处理方法，基于全光谱波长建立了偏最小二乘回归及主成分回归模型，同时，为了提高模型的准确性和稳定性，利用β系数方法选择最优波段1 130～1 147，1 412～1 468，1 658～1 709，2 010～2 055，2 122～2 178和2 395～2 423 nm为自变量建立与双氯芬酸钠浓度的多元线性回归模型。比较了偏最小二乘回归、主成分回归及多元线性回归三种模型。分析结果显示，经标准正态变量预处理方法建立多元线性回归模型，该模型的准确性和和预测能力较好，其预测最低限为0.05%，预测值与实测值的R2为0.992 5，预测均方根误差为0.004 9，标准预测偏差为0.004 9，说明近红外高光谱成像技术为快速定量分析缓解炎症保健食品中违禁添加抗炎药物双氯芬酸钠提供了理论基础。有望进一步开发拓展到其他违禁添加药物的快速定量应用。

A Study on the Screening of Anti-Inflammatory Drug Diclofenac Sodium in Dietary Supplements by Near Infrared Hyperspectral Imaging

Diclofenac sodium is a non-steroidal anti-inflammatory drug with good anti-inflammation effect.The addition of diclofenac sodium is prohibited from being added to the dietary as supplements for reducing the inflammation.At present,some literature has reported that diclofenac sodium in water,meat,milk and other Chinese drugs can be quantified by using high performance liquid chromatography,surface enhanced Raman and electrochemical methods;however,all the mentioned techniques normally require a relatively complex pretreatment process,complicated operation steps and time-consuming.It is necessary to develop a rapid and nondestructive method for determing diclofenac sodium in dietary supplements.This article developeda near-infrared hyperspectral imaging techniqueat 1000~2524 nm combined with chemometrics analysis method for determing diclofenac sodium concentration in the dietary supplements.In this study,eight spectral pretreatment methods were established by using partial least squares regression(PLSR)and principal component regression models(PCR)based on the full spectral wavelength.Furthermore,βcoefficients were used to select the optimal bands in order to improve the accuracy and stability of the model.The multiple linear regression model(MLR)of diclofenac sodium was established at 1130~1147,1412~1468,1658~1709,2010~2055,2122~2178,2395~2423 nm as the independent variable.Subsequently,the three models were compared,and results showed that multiple linear regression model established by standard normal variables pretreatment method had the best forecasting ability.The minimum predictive value of the model accuracy was 0.05%;the Pearson coefficient R 2 of the predicted values was 0.9925;the root means square error of prediction was 0.0049,and the standard deviation of prediction was 0.0049.Therefore,the study has developed the near-infrared hyperspectral imaging technique for rapid determination of diclofenac sodium as an anti-inflammatory drug inthe dietary supplements provides a theoretical basis.It is expected to further developed and extended to the rapid quantitative application of other prohibited added drugs.