论文部分内容阅读
目的:建立快速判别注射用炎琥宁冻干骨架结构的对向传播人工神经网络(counter-propagation artificial neural net-work,CP-ANN)近红外漫反射光谱法。方法:采用配有积分球附件的Antaris II傅里叶变换近红外光谱仪测定注射用炎琥宁的近红外漫反射光谱;用TQ Analyst 8.0进行光谱处理及数据预处理;用注射用炎琥宁校正样品的近红外漫反射光谱数据、以Matlab 6.5建立未损与已损冻干骨架结构的CP-ANN判别模型,并对模型进行交叉验证;用所建CP-ANN模型预测注射用炎琥宁验证样品冻干骨架结构的完整性。结果:所建近红外漫反射光谱CP-ANN判别模型预测注射用炎琥宁冻干骨架结构的准确率为100.0%,且具有良好的可视化功能。结论:所建方法判断客观,无损、无污染,简便快速,可望用于冻干注射用制剂的生产过程质量控制或临床使用质量控制。
OBJECTIVE: To establish counter-propagation artificial neural net-work (CP-ANN) near-infrared diffuse reflectance spectroscopy for the rapid determination of the skeleton of mesylate injections. Methods: Near-infrared diffuse reflectance spectra of memantine for injection were determined by using an Antaris II Fourier transform near-infrared spectrometer equipped with an integrating sphere. Spectroscopy and data preprocessing were performed with TQ Analyst 8.0. The near-infrared diffuse reflectance spectroscopy data of samples were used to establish the CP-ANN discriminant model of undistorted and desiccated freeze-dried skeletal structure with Matlab 6.5, and the model was cross-validated. The CP-ANN model was used to predict Sample freeze-dried skeleton structure integrity. Results: The CP-ANN discriminant model of near-infrared diffuse reflectance spectroscopy was used to predict the freeze-dried skeleton structure of memantine for injection with 100.0% accuracy and good visualization function. Conclusion: The established method is objective, non-destructive, non-polluting, simple and rapid, and is expected to be used in the quality control or clinical quality control of the production process of lyophilized injection preparations.