基于近红外技术的金银花药材多指标成分快速检测

目的 采用近红外光谱(NIR)漫反射法和化学计量学方法对金银花药材质控指标绿原酸和木犀草苷的含量同时进行快速检测，实现金银花药材质量的高效评价。方法 HPLC分别测定不同批次金银花药材中绿原酸和木犀草苷成分的含量，同时采集药材的NIR光谱，最终采用偏最小二乘回归(PLSR)法分别建立NIR与绿原酸和木犀草苷含量间的定量校正模型，并对预测集样品中指标成分的含量进行预测。结果 所建PLSR定量模型的相关系数均>0.90，RMSECV分别为0.469和0.012 4；采用独立验证集对模型进行外部验证，RMSEP分别为0.478和0.010 5，与RMSEC值相近且与RMSECV的比值接近于1，RSEP分别为10.19%和14.76%。所建NIR快速检测方法的准确度和精密度结果良好。结论 该分析方法简便高效，可应用于不同批次金银花药材中绿原酸和木犀草苷2个关键质控指标含量的快速检测及质量控制，结果准确可靠。

Rapid Quantitative Determination of Multi-Indicators in Lonicerae Japonicae Flos by Near-Infrared Spectroscopy

OBJECTIVE To establish a set of quantitative calibration models based on near-infrared (NIR) spectroscopy and chemometrics methods for rapid determination of chlorogenic acid and luteoloside content in Lonicerae Japonicae Flos at the same time, which is useful for the quality control process. METHODS The PLSR algorithm was finally used to build NIR quantitative models between the spectra and the content of chlorogenic acid and luteoloside determined by the reference methods. The validation set was used to identify the predictive performance of the established models. RESULTS For PLSR models, the correlation coefficients of calibration and cross-validation were both >0.90, and the root mean square errors of cross validation (RMSECV) of two indicators were 0.469 and 0.012 4, respectively. External validation proved that the root mean square errors of prediction (RMSEP) were 0.478 and 0.010 5, which were close to RMSEC values, respectively. The ratio of RMSECV to RMSEP was close to 1, and the relative standard errors (RSEP) for chlorogenic acid and luteoloside content were 10.19% and 14.76%. This established method was initiatively proved to be accurate and stable. CONCLUSION NIR can be used for the rapid determination of chlorogenic acid and luteoloside content in Lonicerae Japonicae Flos at the same time, which is beneficial and crucial for the quality control of this raw material in TCM pharmaceutical factory.