ATR-FTIR和UV-Vis结合数据融合策略鉴别滇黄精产地

doi:10.3964/j.issn.1000-0593(2021)05-1410-07

摘要
参考文献
相关文章 (15)

全文: PDF (3645 KB)
输出: BibTeX | EndNote (RIS)

摘要：黄精药材品质优劣与基原植物产地环境因子密切相关，建立简单、快速且能够准确鉴别药材产地的方法对保证其质量可控及用药安全具有重要的理论意义和应用前景。研究中以云南、四川和广西9个产地的133份滇黄精Polygonatum kingianum coll. et Hemsl根茎为试验材料，采集衰减全反射-傅里叶变换红外光谱（ATR-FTIR）和紫外-可见光光谱（UV-Vis）数据预处理后分别建立单一光谱随机森林（Random forest, RF）模型；将ATR-FTIR与UV-Vis数据直接串联完成低级融合，提取两种光谱的主成分数（PCs）和潜在变量（LVs）以实现中级（中级融合_PCs和中级融合_LVs）和高级数据融合（高级融合_PCs和高级融合_LVs），基于不同数据融合策略分别建立RF模型；比较不同模型的正确率（ACC）、灵敏度（SEN）和特异性（SPE），筛选产地鉴别最佳模型。结果显示，不同产地滇黄精ATR-FTIR和UV-Vis峰型相似，吸光度略有差异，ATR-FTIR显示14个共有峰，与糖类、甾体皂苷、黄酮类和生物碱类物质有关，其UV-Vis共有峰主要位于272及327 nm处，与黄酮类物质有关；ATR-FTIR、UV-Vis和低级融合的RF模型，训练集和预测集ACC分别为（76.34%，95.00%），（80.65%，95.00%）和（83.87%，100.00%），但SEN和SPE值较低，故不宜采用；中级融合_PCs和中级融合_LVs的RF模型的SEN和SPE分别为大于0.91和0.98，训练集ACC分别为91.40%和97.85%，预测集ACC均为97.50%；高级融合_PCs和高级融合_LVs的RF训练集ACC分别为77.42%和97.85%，预测集ACC均为95.00%，高级融合_PCs的RF模型鉴别效果较差，高级融合_LVs的RF模型存在过拟合现象；模型鉴别能力为中级融合_LVs＞中级融合_PCs＞低级融合＞ UV-Vis＞ATR-FTIR＞高级融合_PCs；提取LVs对产地鉴别的方法优于PCs；中级融合_LVs建立的RF模型鉴别ACC最高，SEN和SPE大于0.98，模型性能最佳。该方法可为黄精药用资源的科学评价提供理论依据和技术支撑。

关键词：滇黄精；产地鉴别；数据融合；数据衰减全反射-傅里叶变换红外光谱；紫外-可见吸收光谱(UV-Vis)

Abstract：The quality of Polygonati Rhizoma medicinal materials is closely related to the original plants’ origin environment. It is necessary to ensure their quality control and drug safety by establishing a simple, rapid and accurate origin identification method for the medicinal materials.In this study, the attenuated total Reflection-Fourier transform infrared (ATR-FTIR) spectra and ultraviolet visible (UV-Vis) spectra of 133 Polygonatum kingianum rhizomes from 9 geographic origins in Yunnan, Sichuan and Guangxi Provinces were collected to establish random forest (RF) modelafter data pretreatment, respectively. ATR-FTIR and UV-Vis spectra data were directly connected in series to complete the RF model of low-level data fusion. Principal components (PCs) and latent variables (LVs) of the two spectra were extracted to achieve RF model ofmid-level (mid-PCs and mid-LVs) and high-level (high-PCs and high-LVs) data fusion. The accuracy (ACC), sensitivity (SEN) and specificity (SPE) of different models were compared to select the best model for origin identification. The results showed that the peaks of ATR-FTIR and UV-Visspectrain P. kingianum were similar, and their absorbance were different. There were 14 common peaks in ATR-FTIR spectra of P. kingianum, which were related to carbohydrate, steroidal saponins, flavonoids and alkaloids. The common peaks of UV-Visspectra in P. kingianum were mainly at 272 and 327 nm, which were related to flavonoids. For the RF models of ATR-FTIR, UV-Vis and low-level fusion, the ACC of the training set and prediction set were respectively (76.34%, 95.00%), (80.65%, 95.00%) and (83.87%, 100.00%), however, the SEN and SPE values were so low that they were not suitable to use. The SEN and SPE of mid-PCs and mid-LVs RF models were greater than 0.91 and 0.98, respectively. The ACC of the training set was 91.40% and 97.85%, respectively, and that of the prediction set both were 97.50%. The ACC of RF training set with high-PCs and high-LVs was 77.42% and 97.85%, respectively, and the prediction set ACC both were 95.00%. The RF model with high-PCs has poor identification effect, and the RF model with high-LVs was over-fitted. In summary, the identification of model from high to low was: mid-LVs＞mid-PCs＞low fusion＞UV-Vis＞ATR-FTIR＞high-PCs. LVs extraction method is better than PCs for origin identification. RF model of mid-LVs established has the highest ACC with the best model performance, and the SEN and SPE greater than 0.98, and, which can provide a theoretical basis for the scientific evaluation of medicinal resources of Polygonati Rhizoma.

Key words：Polygonatum kingianum; Origin identification; Data fusion; ATR-FTIR; UV-Vis

收稿日期: 2020-05-03 修订日期: 2020-08-12

中图分类号:

O657.3

基金资助: 云南省重大科技专项（2018ZF010），云南省科技计划项目（2017RA001），中医药公共卫生服务补助专项国家重大项目“云南中药资源普查”第5批项目资助

通讯作者: 张金渝 E-mail: jyzhang2008@126.com

作者简介: 张娇，女，1994年生，云南中医药大学中药学院硕士研究生 e-mail：jzhang2019@126.com

引用本文:

张娇，王元忠，杨维泽，张金渝. ATR-FTIR和UV-Vis结合数据融合策略鉴别滇黄精产地[J]. 光谱学与光谱分析, 2021, 41(05): 1410-1416.
ZHANG Jiao, WANG Yuan-zhong, YANG Wei-ze, ZHANG Jin-yu. Data Fusion of ATR-FTIR and UV-Vis Spectra to Identify the Origin of Polygonatum Kingianum. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2021, 41(05): 1410-1416.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2021)05-1410-07 或 https://www.gpxygpfx.com/CN/Y2021/V41/I05/1410

[1] Zhao P, Zhao C, Li X, et al. Journal of Ethnopharmacology, 2018, 214: 274.
[2] Chinese Pharmacopoeia Commission(国家药典委员会). Pharmacopoeia of the People’s Republic of China(中华人民共和国药典), Part One(第一部). Beijing: China Medical Science Press(北京: 中国医药科学出版社), 2015. 306.
[3] ZHANG Jiao, WANG Yuan-zhong, YANG Wei-ze, et al(张娇, 王元忠, 杨维泽, 等). China Journal of Chinese Materia Medica(中国中药杂志), 2019, 44(10): 1989.
[4] LI Jing, WANG Ying-zhe, LIU Yu-cui, et al(李婧, 王英哲, 刘玉翠, 等). China Journal of Chinese Materia Medica(中国中药杂志), 2019, 44(24): 5368.
[5] JIAO Ji, CHEN Li-ming, SUN Rui-ze, et al(焦劼,陈黎明, 孙瑞泽，等). Journal of Chinese Medicinal Materials(中药材), 2016, 39(3): 519.
[6] CHEN Long-sheng, DU Li-ji, CHEN Shi-jin, et al(陈龙胜, 杜李继, 陈世金, 等). Journal of Chinese Medicinal Materials(中药材), 2018, 41(4): 894.
[7] Zhou Y H, Zuo Z T, Xu F R, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2020, 226: 117619.
[8] Zhao Y L, Yuan T J, Zhang J, et al. Journal of Chemometrics, 2019, 33(4): e3115.
[9] Yao S, Li T, Li J Q, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2018, 198: 257.
[10] Wu X M, Zhang Q Z, Wang Y Z. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2018, 205: 479.
[11] Qiu S, Wang J. Food Chemistry, 2017, 230: 208.
[12] Wang Y, Zuo Z T, Huang H Y, et al. Royal Society Open Science, 2019, 6(5): 190399.
[13] Hou L, Liu Y, Wei A. Industrial Crops and Products, 2019, 134: 146.
[14] Rodríguez S D, Rolandelli G, Buera M P. Food Chemistry, 2019, 274: 392.
[15] Chen H, Tan C, Lin Z, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2018, 189: 183.
[16] SUN Yu-qing, HU Yu-zhu, DU Ying-xiang, et al(孙毓庆, 胡育筑, 杜迎翔, 等). Analytical Chemistry(分析化学). 3rd edition(第3版). Beijing: Science Press(北京: 科学出版社), 2011. 136.
[17] Pei Y F, Wu L H, Zhang Q Z, et al. Analytical Methods, 2019, 11(1): 113.
[18] Yang Y G, Zhao Y L, Zuo Z T, et al. Journal of AOAC International, 2019, 102(2): 457.