Near Infrared Spectroscopy Research And Application In Determination Of Antituberculosis Drug

Near infrared spectroscopy technology (NIRS) is a novel technology applying to rapid de-termination of single component or multi-components in materials depend on their opticalcharacteristic in near infrared spectroscopy region. As the NIR is very complicate and seriousoverlap, it is difficult to quantitative analyze multi-components sample depend on the height ofpeaks in spectra, it is necessary to apply a chemometric to develop a multiple regression quanti-tative model using for analysis and prediction. Partial least squares (PLS) is most mentionedand the most efficacious method. This method can extract the efficacious messages, removethe noise and resolve the problem of the collinear in spectra. With its advantages of no pre-treatment, no pollution, facility, non-destructibility and online measurement, it has been gener-alized to on-line and real-time quality control in pharmacy.In this paper, I have studied on the application of NIRS on rifampicin, isoniazid and com-pound antituberculosis drug. And then I studied on the influence of pretreatment methods ofNIR spectra, such as Savitzky-Golay smoothing, Fast Fourier Transform (FFT) and derivativeon establishing quantitative analysis model, discussed the optimal wavelength region and op-portune factors influences in the models.1. Rifampicin capsule: Applied data pretreatment approaches includingSavitzky-Golay smoothing, FFT, derivative to dispose the NIR spectra of rifampicin, and de-veloped NIRS-PLS models for determination of rifampicn and excipients in rifampicin capsules.I have constructed. PLS models for determination of rifampicin and excipients in rifampicintablets. In these models, the model for determination of rifampicin was using Savitzky-Golaysmoothing spectra in the range of 1416~2252 nm, its fit number of factors is 7 and its RootMean Square Error of Cross-Validation (RMSECV) is 0.00651 , the model for determinationof starch was using FFT spectra in the range of 1100~1416 nm , its fit number of factors is 6and its RMSECV is 0.00229, the model for determination of carboxymethyl cellulose wasusing FFT spectra in the range of 1100~1416 nm , its fit number of factors is 7 and itsRMSECV is 0.00182 , the model for determination of dextrin was using FFT spectra in therange of 1416~2252 nm , its fit number of factors is 7 and its RMSECV is 0.00207. Applyingthese models to determination of rifampicin, starch, catboxymethyl cellulose and dextrin inrifampicin capsule, their RMSEP are 0.00380, 0.00116, 0.00122 and 0.00123. This method hasgood repeatability, the RSD of 25 times scan of the same sample for determination of rifampicin,starch, catboxymethyl cellulose and dextrin are 0.858%, 0.708%, 1.49% and 0.516%.2. Isonazid tablet: Applied data pretreatment approaches according to Rifampicincapsule to dispose the NIR spectra of isoniazid, and developed NIRS-PLS models for determi-nation of isoniazid in isoniazid tablets. Its best pretreating method is FFT method, the effica-cious spectra region is 1323~1540 nm, and the fit number of factors is 7, its RMSECV is0.00515. Using this model to predict the medicine amounts in validation set, the RMSEP is0.00614. The recovery of this method in validation set is 100.298%. RSD of recovery determi-nation of the same sample (13 times) is 0.664%. RSD of reappearance (25 times) is 1.475%.These results indicate that this method is accurate, precise and repeatable.3. Compound antituberculosis drug: Applied data pretreatment approaches ac-cording to Rifampicin capsule to dispose the raw NIR spectra, and developed NIRS-PLS mod-els for determination of rifampicin, isoniazid and pyrazinamide in compound antituberculosisdrug. The model for determination of rifampicin using first derivative spectra in the range of1100~1853 nm, the length to the point is 50, its fit number of the factors 4 and its RMSECV is0.00287. The model for determination of isoniazid using first derivative spectra in the range of1100~1853 nm, the length to the point is 10, its fit number of the factors 6 and its RMSECV is0.00216. The model for determination of pyrazinamide using FFT spectra in the range of1100~1853 nm, the length to the point is 10, its fit number of the factors 6 and its RMSECV is0.00216. Applying these models for predicted the rifampicin, isoniazid and pyrazinamidecontents in prediction set, the RMSEP are 0.00298, 0.00277 and 0.00304 and their average re-covery are 100.708%, 100.292% and 99.803%.In this paper, I studied on the pretreatment of NIR spectra, applied the PLS to construct aquantitative determination model to predict the amount of rifampicin, isoniazid, and compoundantituberculosis drug. The results indicate that it is feasible to apply near infrared spectroscopycombine with PLS to non-destruction determination of single or compound antituberculosisdrug. Precision of this method is up to scratch in the produce process of antituberculosis drug. Itcan be generalized to on-line and real-time quality control in pharmacy.