Quantitative Determination Of The Active Ingredient In Medicine Using Near Infrared Diffuse Reflection Spectroscopy Combined With Multivariate Calibration

There are six parts in this thesis. Near infrared spectroscopy combined with chemometrics methods and its application were reviewed. The multivariate calibration methods used in this dissertation and their principles were also introduced. The quantitative analysis of ferulic acid in Ligusticum chuanxiong Hort, a new antiepileptic drug topiramate, and anti-hypertension drug captopril using these metod combined with NIR spectroscopy were investagated.The results demonstrated satisfactory.In the first chapter, the principle, features of near infrared spectroscopy, quantitative analysis method and their application in the field of drug analysis of NIR spectroscopy combined with chemometrics methods are introduced. The second chapter is the presentation of the chemometrics methods which are used in this dissertation.In the third chapter, near infrared (NIR) spectroscopy and chemometrics were applied to the quantitative analysis of ferulic acid in Ligusticum chuanxiong Hort. Partial Least Squares (PLS) regression model was built using a calibration set (34 samples). The prediction set was established by 11 samples. And we applied this method to predict the data in the prediction set. The correlation coefficient (R) of PLS model was 0.9883 as indicated from the cross validation, and the root mean square error of cross validation (RMESCV) was 0.146%. The correlation coefficient (R) of the reference and predicted values was 0.9751, and the root mean square of prediction (RMSEP) was 0.251%. NIR spectroscopy method was simple, rapid and accurate. The method could be applied in rapid qualitative and quantitative of different batches of Ligusticum chuanxiong Hort from Sichuan province.In the fourth chapter, near infrared (NIR) diffuse reflectance spectroscopy combined with partial least squares (PLS) was attempted to analyze the topiramate content in topiramate powder. The spectral data of topiramate were obtained though NIR spectrophotometer and preprocessed. Then, partial least squares regression model was set up using a calibration set (39 samples). The prediction set was set up by 13 samples. And we applied this method to predict the data in the prediction set. The result showed that the correlation coefficient (R) of PLS model was 0.9345 as indicated from the cross validation, and the root mean square error of cross validation (RMSECV) was 0.0169. The correlation coefficient (R) of the reference and predicted values was 0.9155, and the root mean square of prediction (RMSEP) was 0.0239. NIR spectroscopy method was simple, rapid and accurate. This method has significant potential in rapid qualitative and quantitative of different batches of topiramate.In the fifth chapter, principal component analysis-back propagation neural network algorithm (PCA-BPNN) model of topiramate was established using chemometric methods combined with near infrared spectroscopy. And the effect of input nodes, hidden nodes, epochs, learning rate and momentum on BPNN model was investigated in order to acquire the optimal calibration model. The result showed that the PCA-BPNN model of topiramate was quite stability, and its predicted values were accurate and reliable. This method was simple and rapid, and it provides a reference method for testing the drug.In the sixth chapter, the properties of PLS model and PCA-BPNN model of captopril component were compared. Results were followed, both PLS model and PCA-BPNN model of captopril component were able to satisfy the requirements of error range in the actual production. As the PCA-BPNN method is a relatively new analytical method that development is still not perfect in certain details of the operation. So, PCA-BPNN is a suitable tool to model the non-linear complex system, with additional advantages over PLS.