Research Of Passive Shimming Algorithms For The Superconductiong MRI Systems

Magnetic resonance imaging (MRI) technology has developed into an important non-intrusive tomographic imaging technique, and is playing a more and more important role in clinical medical diagnosis and treatment. Because of high magnetic field, high time stability and high spatial uniformity, superconducting MRI systems have become the mainstream in modern clinical application, they also have bright prospects in science research and other aspects. The imaging of a superconducting MRI system and its image quality is heavily dependent on the main magnet which produces the uniform static magnetic field of high strength within the imaging volume. However, because of the field errors caused by the process of the designing, manufacturing and installation of superconducting magnets, the uniformity of the main magnet has to be corrected to reach the requirements of producing MRI images of high quality. There are usually two techniques to do the shimming of MRI systems, one of them is active shimming technique which uses coils carrying current to improve the uniformity of the magnetic field within the imaging volume but these coils usually compete for available space with gradient coils, the shimming coils are also of high cost. The other technique is passive shimming, passive shimming technique optimizes the layout of a series of shims (magnetic material, usually steel) with proper size to improve the uniformity of the static field and once applied it doesn’t need power. Passive shimming is very cheap and flexible. The shim pieces are fixed in the drawers in the inner bore of the superconducting magnet, and produce magnetizing field in the imaging region to compensate for the inhomogeneity of the B0 field. In practice, the total mass of steel used for shimming should be minimized in addition to the field uniformity requirement within the volume of interest, because the presence of steel shims will introduce thermal stability problem and the problem of eddy currents. In order to correct the main field inhomogeneity of superconducting MRI systems by passive shimming, the research work of this thesis includes the following three aspects.(1) A novel passive shimming method for the correction of magnetic fields in magnetic resonance imaging systems.Passive shimming is used to find an optimum configuration for the placement of iron pieces applied to improve the Bo uniformity in the predefined imaging region referred to as the diameter of spherical volume (DSV). However, most passive shimming methods neglect to recognize that the subspace of the imaging volume under the patient bed is not in use for imaging. In this work, we present a novel shimming method that attempts to avoid the unnecessary shimming of the space under patient bed. During implementation, the Bo field is still measured over the DSV surface and then mapped onto the effective imaging volume surface which is the subspace of the DSV above the patient bed; a dedicated sensitivity matrix is generated only for the imaging area above the bed. A linear programming optimization procedure is performed for the determination of thicknesses and locations the shim pieces to shim the magnetic field. Our experimental results showed that by revising the shimming target area, the new method provides superior optimization solutions. Compared to a conventional approach, the new method requires smaller amount of iron to correct the Bo inhomogeneity in the imaging area which has the effect of improving thermal stability to the Bo field and decrease the eddy currents. It also reduces the complexity of the optimization problem, saves resources. Our new shimming strategy helps to improve the magnetic field homogeneity within the realistic imaging space of magnetic resonance imaging systems, and ultimately improve image quality.(2) Passive shimming using the L1-norm regularized least square algorithm.The passive shimming procedure is typically realized using the linear programming (LP) algorithm. But in practical shimming process, the LP approach however, is generally slow and also has difficulty balancing the field quality and the total amount of shims for shimming. For this, in this paper, we have developed a new algorithm that is better able to balance the dual constraints of field uniformity and the total mass of the shims. The least square method is used to minimize the magnetic field inhomogeneity of the sample points over the imaging surface with the total mass of steel being controlled by an Li-norm based constraint. The proposed algorithm has been tested with practical field data, and the results show that, with similar computational cost and mass of shim material, the new algorithm achieves superior field uniformity (43% better for the test case) compared with the conventional linear programming approach. At the same time, this algorithm overcomes the defect of long running time of usual nonlinear algorithms used in passive shimming. When both of the amount of the drawers of shims and the sample points reach 500, the one-time running of this algorithm is within half an hour, if using the initial solution from LP algorithm it will be within minutes, so this algorithm can provide optimal results rapidly.(3) Particle Swarm Optimization algorithm used in passive shimming.As a key technology of improving the magnetic field uniformity within the imaging volume, passive shimming is widely used in the shimming of MRI magnets because of its low price and great flexibility. However, since the number of the variables to be optimized in the passive shimming optimization reaches several hundred even thousand, it is a high dimensional optimization, general deterministic algorithm will have difficulty in further improving the uniformity after it has been shimmed to some degree of uniformity when shimming the main magnetic field. This thesis thinks about to do the second optimization for main magnetic field of the MRI systems using the PSO algorithm, an artificial intelligence. We planned to do further improve the field uniformity after it was first optimized by the traditional algorithm, the linear programming algorithm, using the PSO method as a fine tuning to acquire MR images of higher quality. The 5th chapter of this work simulated the above method on the basis of traditional LP algorithm, it proved that the PSO could slightly further improve the field uniformity after it was first optimized by LP algorithm. Because the uniformity of static magnetic field in the imaging volume of high field superconducting MRI systems usually reaches 10 ppm (of very small magnitude) after the LP optimization, the common parameters in traditional PSO should make corresponding adjustment to further improve the final results. Moreover, we find that the optimal results of the particle swarm optimization have relation with the number of the initial particles and their diversity, and this algorithm will easily run into local optima. The application of particle swarm optimization in passive shimming of magnetic resonance imaging systems need more research work and improvement.