Optimisation of passive shimming techniques for magnetic-resonance imaging

The aim of this work was to develop and optimise techniques for improving the magnetic-field homogeneity of magnetic-resonance imaging (MRI) systems using ferromagnetic material (passive shimming). The research was carried out as a collaborative CASE Award with Magnex Scientific, with the emphasis on assessing the performance of two algorithms in finding the required spatial distributions of ferromagnetic material to annul unwanted variations a given magnetic field. The work was carried out on a 2 Tesla MRI system (310 mm horizontal-bore superconducting magnet), with the intention of making it capable of performing in-vivo spectroscopy, chemical-shift imaging and high-resolution skin imaging. The results of this work, however, have applications to any MRI system - particularly systems targeted to image specific anatomy e.g. the wrist or knee. Magnetic field measurement and passive shim systems were designed and built to operate within the physical limitations of the narrow bore of the magnet used in this research. The two algorithms used were the 'quasi -continuous' algorithm (an analytic target-field approach) and the 'discrete' algorithm (an iterative passive shimming method designed to converge on an acceptable solution). The discrete algorithm performed best, improving the plot peak-to-peak inhomogeneity from 16.7 #+-# 0.8 parts-per-million to 4.4 #+-# 0.5 parts-per-million (across a 96 mm diameter sphere centred on the magnet isocentre) using only 144 g of steel. The quasi-continuous algorithm improved the field to 7.4 #+-# 0.6 parts-per-million across the same volume using 142 g of steel. The shimming of the magnetic field is very sensitive to errors within the system. In general, the shimming process requires many iterations to attain the required magnetic-field homogeneity, thus making it a time consuming process. A method is described for evaluating the errors on the harmonic analysis of the field, using an approximation based on the local field-gradient at the measurement point. This method does not require the user to perform repeat measurements of the B-field, and thus has applications for engineers in the field when setting up clinical or research imaging systems, since repeat measurements are not generally possible due to the limited amount of time available for this procedure. As part of this evaluation, an alternative definition of inhomogeneity has been devised. This definition is less sensitive to field measurement errors than the standard 'plot peak-to-peak' inhomogeneity, and is insensitive to the positions of the magnetic-field measurement points. The new definition also allows determination of the dominant source of error on the shim system, which has applications to diagnosing problems in any field-measurement or shim system. To demonstrate the performance of the 2 Tesla MRI system following passive shimming, fat-suppression techniques were performed on the knuckle joints of human volunteers resulting in 93% suppression of the fat signal. Images taken with this technique show an improved definition of the cartilage layer and could be used to monitor the progression of diseases such as rheumatoid arthritis, where fatty infiltration of joint structures during active disease is an important determinant of therapeutic strategies. (author)Available from British Library Document Supply Centre-DSC:DXN028602 / BLDSC - British Library Document Supply CentreSIGLEGBUnited Kingdo