Magnetic field gradients are a fundamental requirement for magnetic resonance imaging (MRI). Consequently, as MRI techniques have become more advanced in recent years, the demands on the hardware producing these fields have increased. This thesis aims to demonstrate a novel method for designing gradient coils using a genetic algorithm (GA). The technique is shown to be both powerful and flexible, allowing incorporation of many practical considerations directly into the optimisation procedure and even facilitating designs based on unconventional geometries, with both flat and dual-diameter coils studied. The optimisation itself is written to favour robust coil designs and constraints are specified to ensure that all designs are relatively simple and thus cheap to construct. The effect on the achieved outcome of individual intrinsic and implementation-specific variable design parameters is examined. The manufacture, testing, evaluation and clinical application of two gradient coils designed using the GA are discussed in the contexts of human brain and knee imaging. The former is accepted as a particularly stringent test of gradient performance, and artefacts due to the inevitable imperfections of the magnetic field produced by such a gradient coil are largely removed using an algorithm compatible with the design software. An attempt is made to compare the results obtained with data available from the literature, and this leads to discussion of the difficulties in assessing and presenting measures of the quality of gradient hardware, but also shows that the designs are competitive with those produced by other strategies. Subsequent to acquiring high quality images of the knee at full flexion, the advantages of flat gradient coils are further made apparent by a number of computer simulations which indicate their potential for the future of MRI. The advances made during this thesis have made the GA technique the method of choice locally, and have enabled several major imaging studies.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:599045 |
| Date | January 1997 |
| Creators | Fisher, B. J. |
| Publisher | University of Cambridge |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
Page generated in 0.0018 seconds