Chemometric Algorithms In The Study Of Traditional Chinese Medicine Fingerprint

With the development of analytical techniques and desirable progress of relevant researches about traditional herbal medicines, both the compilations of pharmaceutical attributes and the comparison of componential distributions of herbal samples have been becoming more and more important for the medicinal modernization, industrialization, internationalization, in particular quality assessment, to support their uses. With the help of modern hyphenated instruments and related chemometric methods, it is possible to deal with the complicated systems. This research aims at the application of chemometric methods to the analysis for group substances in real herbal medicines in order to quality control.In the first chapter of this thesis, the development for quality control of herbal medicines was summarized. The application of chromatographic fingerprints, modern analytical techniques, and chemometric methods are very useful to the quality control of herbal medicines.In chapter 2, multicomponent spectral correlative chromatography was proposed as a new method based on orthogonal projection and correlative chromatogram. In order to eliminate the influence from heteroscedastic noise, the cosine value between target component spectrum and its projected residue was applied to detect spectralcorrelative component chromatographic clusters of different fingerprints of herbal medicines. It could practically facilitate qualitative evaluation of herbal medicines.In chapter 3, the chromatographic fingerprint of Erigeron breviscapus herbal sample was investigated and established by using many extract methods and optimizing experimental conditions. The result was important for the further study to the chemical analysis and quality control of Erigeron breviscapus.In chapter 4, data pretreatment, local least squares and principle component analysis were applied to chemical pattern recognition for qualitatively analyzing the chromatographic fingerprints of various herbal samples of Erigeron breviscapus. The classification of them demonstrated that most of the tested samples have good similarities with the common pattern fingerprint of the regular samples and some were inferior or even spurious, appearing apart from the regular samples at the project position in principle component analysis. This pattern recognition was also a good way to the identification and evaluation of Erigeron breviscapus and might be of practical value for controlling the quality of traditional herbal medicines.