| The use of Magnetic Resonance Imaging (MRI) for medical diagnosis has grown tremendously over the past two decades. However, MRI scans are expensive with typical scan costs exceeding {dollar}1,000. Prepolarized MRI (PMRI) is a possible low-cost alternative to conventional MRI that uses dedicated scanners for imaging specific body parts such as head, neck, and extremities. MRI requires a strong and homogeneous magnetic field which is achieved using an expensive superconducting magnet. PMRI uses two inexpensive resistive magnets instead, one strong but inhomogeneous, and the other weak but homogeneous. A typical MRI scanner costs over {dollar}1,000,000. A PMRI scanner with image quality equivalent to a 0.5 Tesla conventional scanner will be less than {dollar}50,000 in capital costs.; Since PMRI scanners are designed to fit specific body parts and since PMRI works with two resistive magnets, integration is a challenge. The main system components are the strong polarizing magnet, the homogeneous readout magnet, the gradient coils, the RF and receiver coils, and the console. Following a brief discussion on the general working principles of MRI and PMRI, this thesis presents my contributions to the three projects concerning the design of the readout magnet and the gradient coils.; A homogeneous readout magnet design algorithm based on linear programming was developed. It handles arbitrary geometric constraints and its flexibility allows efficient system integration. An extremity readout magnet designed using this algorithm was constructed and tested.; The fast speed of this algorithm allowed us to design hundreds of magnets with different geometries. As a result, we discovered a simple linear relation between a cylindrical magnet's length, bore size, and homogeneous volume dimensions.; Finally, we developed an algorithm for the design and optimization of the gradient coil. The underlying principle was similar to the homogeneous magnet design algorithm, but significant modifications were made to allow for two dimensional current distributions on arbitrary surfaces. A gradient coil set designed using this algorithm was built and successfully tested. |
