不同近红外光谱仪用于酒石酸美托洛尔片厂家鉴别的对比研究

Identification of Metoprolol Tartrate Tablets from Different Manufacturers by Different Near Infrared Spectrometers

建立近红外光谱定性分析模型,用于酒石酸美托洛尔片厂家的快速鉴别。采用AntarisⅡ傅里叶变换近红外光谱仪和Micro NIR1700型微型近红外光谱仪两种仪器对来自四个生产厂家的66个批次的酒石酸美托洛尔片进行近红外光谱的采集,选用随机（random sampling,RS）法进行样品集的划分,得到校正集样品44个、验证集样品22个,选择一阶导数Savitzky-Golay15（SG15）点平滑、标准正态变量变换（SNV）作为预处理方法,根据不同类样本间光谱的差异结合谱带特征吸收的方法分别选择6 468~7 104和6 468~7 156cm-1作为建模光谱区间,建立两种采样仪器的偏最小二乘判别分析（PLS-DA）模型。模型的混淆矩阵结果显示两种仪器采集的近红外光谱建立的定性分析模型能够快速、有效的鉴别不同厂家的酒石酸美托洛尔片,模型的准确率和否定率均达到100%。该研究有效证实了近红外光谱分析技术用于酒石酸美托洛尔片厂家定性鉴别的可行性,另外,该研究中采用了目前商业化最小、便于携带的Micro NIR1700型微型光谱仪,为药品的快速现场筛查提供了很好的启示。

To rapidly identify Metoprolol Tartrate tablets from different manufacturers ,a qualitative analysis model can be estab‐lished by near infrared spectroscopy .Firstly ,AntarisII Fourier‐transform near infrared （FT‐NIR） spectroscopy and Micro NIR1700 Spectrometer were used to collect spectral data of 66 batches of samples which come from four different manufacturers , then 44 samples of calibration set and 22 samples of validation set were acquired by random sampling .In order to build up a PLS‐DA model ,the first derivative with Savitzky‐Golay 15 points smoothing （1d＋SG15） and standard normal vitiate transformation （SNV） was selected as the pretreatment method and according to the variation between different samples and the characteristic absorption band ,6 468-7 104 cm^-1 and 6 468-7 156 cm ^-1 were chosen as the modeling spectra region .The confusion matrix indicated that Metoprolol Tartrate tablets could be rapidly and effectively identified by two analytical models ,which were estab‐lished using the spectral data collected from two instruments .For these two models ,both of the sensitivity and specificity were 100% .This study confirmed that it is feasible to carry out the manufacturer identification of Metoprolol Tartrate tablets by near infrared spectroscopy .Besides ,the use of Micro NIR1700 Spectrometer ,which is the minimum and portable near infrared spec‐trometer ,provides valuable insights for fast on‐site drug screening .