Study On Automatic Baseline Correction Method For Nuclear Magnetic Resonance Spectra

Nuclear Magnetic Resonance (NMR) technique has greatly improved the structure elucidation of molecules in solution as a noninvasive measurement. It has been widely used in many fields such as chemistry, material, biology and medicine, and plays a more and more important role in these fields. Among the sources of error in most quantitative analysis in NMR spectra are distorted baselines. In order to eliminate the baseline distortion, researchers have proposed a variety of baseline correction methods from different aspects. As two-dimensional and even multidimensional spectra proposed in improving NMR techniques, manual methods for baseline correction are no longer able to handle spectra with large datas. Researches on automatic baseline correction methods become significant. However, data points in peak region are rarely used in the existing automatic baseline correction methods, which can lead to negative regions and a change to peak intensity. As a result, the establishment of a more accurate techniques for NMR automatic baseline correction would be of great help to improve quality of spectra, as well as accuracy and precision of NMR analysis. The main work of this thesis is listed as follows.1. The research background of automatic baseline correction method was briefly introduced firstly. Then a summary of the reasons that caused distorted baselines was presented. A comprehensive review was made about automatic baseline correction methods for NMR spectra, and basic principles of several full-and semi-automatic methods were described respectively, along with their merit and demerit.2. A new automatic baseline correction method based on iterative for NMR spectra was proposed. It is based on an improved baseline recognition of peak regions using iterative moving averaging algorithm, followed by an iterative baseline modeling procedure. The method combines advantages of two baseline correction methods in order to improve baseline modeling in peak regions. A series of "baseline anchor points" in peak regions are picked during baseline recognition, and utilized to participate in baseline modeling with baseline points. The baseline correction method has been shown to perform well on NMR spectra even for spectra with a large baseline distortion.