文献类型：期刊文献

中文题名：基于傅里叶变换红外光谱技术对中药党参产地快速判别方法的研究

英文题名：Research on Rapid Identification Method of Origin of Dangshen(Codonopsis pilosula)Based on Fourier Transform Infrared Spectroscopy Technology

作者：杜晓伟[1];宋平顺[1,2];倪琳[1,2]

第一作者：杜晓伟

机构：[1]甘肃中医药大学药学院,甘肃兰州730099;[2]甘肃省药品检验研究院,国家中药材及饮片质量控制重点实验室,甘肃省中藏药检验检测技术工程实验室,甘肃兰州730070

第一机构：甘肃中医药大学药学院（西北中藏药协同创新中心办公室）

年份：2025

卷号：43

期号：3

起止页码：75

中文期刊名：中华中医药学刊

外文期刊名：Chinese Archives of Traditional Chinese Medicine

收录：;北大核心:【北大核心2023】；

基金：国家药品标准修订项目(BZ2021275);甘肃省科学监管项目(2021GSMPA-KL001,2021GSMPA-KL002)。

语种：中文

中文关键词：傅里叶变换红外光谱技术;中药;党参;产地;化学计量学

外文关键词：FT-IR;Chinese medicines;Dangshen(Codonopsis pilosula);place of origin;chemometrics

摘要：目的在实际检测中发现,31批不同产地党参的浸出率各不相同,最低和最高相差20.869%,所以即使是正品中药材,也会因为产地不同而有不一样的药效。而这31批党参的外形都十分相似,无法通过性状鉴别对其进行区分;又因为31批党参均为正品,其主要成分相同,无法用显微鉴别和薄层鉴别对其区分;而前人对中药产地判别的方式如液相色谱法、气相色谱法和聚合酶链式反应等方式,虽然准确但需要数小时甚至数天的专业人员检测。所以以党参为例,研究基于傅里叶变换红外光谱技术快速判别中药产地的方法。方法随机选取10批不同产地的党参,分别采集红外光谱数据;首先计算不同产地党参红外光谱的相似度并绘制相似度热图进行初步分析;对原始红外光谱进行5种不同方法预处理,建立偏最小二乘模型并选出最优处理方式,对选出的最优组进行聚类分析;利用连续投影算法选出最优处理方式组的最优波长,建立多光谱模型并计算出产地判别的回归方程;最后利用混淆矩阵对回归方程的判别性能进行验证。结果全光谱模型最优处理方法为归一化处理,10批不同产地的党参经最优处理后的光谱数据可通过聚类分析完全聚为10类,且相同产地的样品均聚在同一类别里;全光谱模型组的预测集决定系数R2 p为0.9800,对应的RMSEP为0.4067%,连续投影算法挑选出的13个最优波长建立的多光谱模型的预测集决定系数R2 p为0.9906,RMSEP为0.2804%,多光谱模型的预测性能优于全光谱模型,且需要输入的变量大大减少,省时省力准确度高;基于多光谱模型计算出的产地判别回归方程通过混淆矩阵的验证,得出其灵敏度和特异性均为97%以上,故认为回归方程可以快速准确地判别党参的产地。结论以中药党参为例研究的傅里叶红外光谱技术判别中药产地的方法快速、有效,为其他中药的产地判别研究提供思路,为后期开发中药快检设备提供理论基础。
Objective In the actual test,it is found that the leaching rate of 31 batches of Dangshen(Codonopsis pilosula)from different origins is different,with the lowest and highest differences of 20.869%.Therefore,even genuine Chinese medicine will have different effects depending on the place of origin.The appearance of these 31 batches of Dangshen(Codonopsis pilosula)is very similar,and it is impossible to distinguish them by trait identification.Because the 31 batches of Dangshen(Codonopsis pilo-sula)are all genuine and their main components are the same,they cannot be distinguished by microscopic identification and thin layer identification.In the past,the methods of judging the origin of traditional Chinese medicines,such as liquid chromatogra-phy,gas chromatography and polymerase chain reaction,were accurate but required professionals to test for hours or even days.Therefore,this paper took Dangshen(Codonopsis pilosula)as an example to study a method to quickly identify the origin of tradi-tional Chinese medicine based on fourier transform infrared spectroscopy technology(FT-IR).Methods Ten batches of Dangsh-en(Codonopsis pilosula)from different origins were randomly selected and infrared spectrum data was collected separately.First-ly,the similarity of the infrared spectra of Dangshen(Codonopsis pilosula)of different origins was calculated,and the similarity heat map was drawn for preliminary analysis.Five different methods were used to preprocess the original infrared spectrum.The partial least squares(PLS1)model was established and the optimal processing method was selected and the cluster analysis was carried on the selected optimal group.The continuous projection algorithm was used to select the optimal wavelength for the opti-mal processing method group,establishing a multispectral model and calculating the regression equation for identifying the pro-duction location.Finally,the discriminant performance of the regression equation was verified by using the confusion matrix.Re-sults The optimal processing method for the full-spectrum model was normalization(NOR).The spectral data of 10 batches of Dangshen(Codonopsis pilosula)from different origins after NOR treatment can be completely clustered into 10 categories by clus-ter analysis,and samples from the same origin were all clustered in the same category.The value of R2 p of the full spectrum model group was 0.9800,the corresponding RMSEP was 0.4067%,and the value of R2 p of the multispectral model established by the 13 optimal wavelengths selected by SPA was 0.9906 and RMSEP was 0.2804%.The prediction performance of the mul-tispectral model was better than that of the full-spectrum model,and the variables that needed to be input were greatly reduced,saving time and effort,and the accuracy was high.The regression equation of origin discrimination calculated based on the multi-spectral model was verified by the confusion matrix,and its sensitivity and specificity were more than 97%,so it is believed that the regression equation can quickly and accurately determine the origin of Dangshen(Codonopsis pilosula).Conclusion The FT-IR method of determining the origin of traditional Chinese medicines studied by taking Dangshen(Codonopsis pilosula)as an ex-ample is fast and effective,which provides ideas for the research on the origin of other traditional Chinese medicines and provides a theoretical basis for the development of rapid inspection equipment for traditional Chinese medicines in the future.