Research And Implementation Of NMR Automatic Methods

Automation is an important trend in the development of modern nuclear magnetic resonance(NMR).The automatic methods of experiment and data processing are of great significance to reduce the repetitive operations of researchers and improve the automation level of instruments.In the research of automatic methods,new ideas are proposed to improve the accuracy and robustness of automatic gradient shimming and phase correction.For the software implementation of automatic methods,more flexible and extendible software architecture is proposed to improve the ability to integrate diverse and complex automatic methods.This thesis is mainly composed of three parts:(1)A phase encoding based gradient shimming method is proposed to improve the effect of gradient shimming in case of poor magnetic field homogeneity.Based on the frequency encoding,a large inhomogeneous magnetic field seriously affects the accuracy of the imaging for the traditional gradient shimming method.There are two reasons:first,the time domain signal decays rapidly because of short T2*;Second,with inhomogeneous magnetic field added to the gradient field,the accuracy of spatial encoding is reduced,and the magnetic field data will be distorted.In this thesis,a new gradient shimming method based on phase encoding is proposed to solve above problems.Phase encoding based gradient shimming eliminates the dispersion and refocusing time of gradient,so the data can be acquired at an earlier time;Undistorted magnetic field data can be acquired even when the homogeneity of magnetic field is poor,because phase encoding is unaffected by the inhomogeneity of magnetic field.In addition,phase encoding is unaffected by chemical shift,it is a new solution to solve the problem of multiple peaks for 1H gradient shimming.(2)A more robust automatic phase correction method based on the segmented optimization of baseline.Existing automatic phase correction methods usually utilize the characteristics of peaks as the evaluation for phase optimization.But it is difficult to achieve ideal result in case of overlapping peaks and inhomogeneous magnetic field.Proposed method combines the smoothness characteristics of baseline and non-negative nature of signals for phase optimization.It uses state-of-the-art baseline fitting algorithm "arPLS"to improve the efficiency and accuracy.When the smoothness of baseline is maximized and negative signals are minimized,the algorithm convergences.Segmented optimization is used to support negative and distorted peaks.(3)A plug-in based software architecture is proposed to implement and integrate diverse and complex automatic methods.To integrate diverse and complex automatic methods,NMR software should be highly flexible and extendible.In this thesis,a plug-in based software architecture is proposed.The construction of plug-in system and extension mechanism are analyzed.Conventional automatic methods with plug-in implementation are discussed.New automatic methods can be easily integrated into the plug-in system without changing other modules.The software provides a flexible and extendible workbench for NMR researchers.