The New Methodologies To The Analysis Of 3D HPLC-PAD Fingerprint Spectra

In recents years, it is easier to obtain the three-dimension (3D) spectra with the development and application of the hyphenated techniques. Compared with the traditional two-dimensional (2D) spectra,3D spectra consisted of the entire characteristics of deteced samples; thus, more accurate and reliable quantitative analyses results will be obtained. The composition of traditional Chinese medicines (TCMs) and their preparations is complex, thus, great effort may be necessary by the application of the traditional methods to optimize the chromatographic conditions (such as mobile phase, gradient condition, and column temperature etc.) to obtain the baseline separation, which extend the analytical time and experimental work. In these cases, one can resort the chemomeric approaches, which could decrease the requirements of the separation and give accurate results. In this dissertation, the new methodologies to the analysis of 3D HPLC-PAD fingerprint spectra have been proposed, and these newly proposed methods not only been fully and strictly validated, but also compared with the reported methods, and further applied to slove the practical problems. These following points are composed in this dissertation.Chapter 1 First of all, a brief overview of the development and application of HPLC, the acquisition and advantages of TCMs fingerprint, and the application of the chemometrics methods for the analysis of fingerprint have been made, respectively. Then, a brief summary of the research works in this dissertation bas been carried out.Chapter 2 Accumulated Projection (AP) method was developed and applied to the determination of four active compounds in real Sanqi Panax Notoginseng samples based on 3D HPLC-PAD spectra, and the performance of the AP method was fully validated. The obtained results demonstrated that the AP method can be applied to the fast quantifying the target compounds in the real samples, even with partial overlapping peaks of target compounds and in the presence of unknown interferences. Compared with the N-PLS method, the proposed method not only easy to understand but also could efficiently obtain the quantitative information of the target components, and presented accurate results.Chapter 3 An efficient approach, longest distance shifting (LDS), was presented to align shifted chromatographic signals. Firstly, the fundamental alignment theory was reviewed. Then, the developed algorithm was evaluated by two data sets of simulated chromatograms that covered the different kinds of peak shifts in retention time, and further confirmed with experimental chromatograms. Compared with correlation optimized warping (COW) and Interval Correlation Optimised Shifting (icoshift) algorithms, LDS algorithm is simpler, faster and can better preserve the shape of peaks and achieve a higher quality of the alignment. The obtained results indicated that the alignment procedure of LDS method is simpler and easier to understand and it is insensitive to the choice of segment boundaries.Chapter 4 A practical application of wavelet moment (a kind of image moment) method on the quantitative analysis of the three main active components (Chlorogenic acid, forsythin and baicalin) in Shuanghuanglian oral liquid samples was presented on the basis of the 3D HPLC-PAD spectra, and the obtained results were fully validated. The results demonstrated that the wavelet moment method with the advantages of multi-resolution and inherently invariance properties in translation, rotation and scaling can be applied to the simultaneous determination of the target components successfully, even with overlapped peaks, retention time shifts and noise signals.Chapter 5 Another kind of image moment method (Tchebichef moment) based on the 3D HPLC-PAD spectra was proposed and used to determine four main active components (notoginsengside R1, ginsenoside Rgl, ginsenoside Re and ginsenoside Rb1) in Sanqi Panax Notoginseng Injection samples. The results demonstrated that the proposed method also can be applied to analyze multiple target compounds based on their 3D HPLC-PAD spectra accuratly, in which there were overlapping peaks, unknown interferences and chromatographic drifts. Besides, the Tchebichef moment is a kind of discrete orthogonality moment, which does not require numerical approximations derived from continuous domain and image coordinate transformations since the moment kernels are polynomials of discrete variables and the domain of polynomials matches the discrete domain of the digital image. Therefore, the Tchebichef moments eliminate the extra numerical approximation and maintain the orthgonality, invariance and multi-resolution properties perfectly, which makes the Tchebichef moments superior to the traditional continuous orthogonal moments in terms of image representation capability, particularly in applications of requiring independent shape characteristics. Furthermore, several moment methods applied in analytical chemistry were compared, which provided a valuable insight into the application of moment methods in the analytical field.