Date of Award

December 2012

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Engineering

First Advisor

Jun Zhang

Second Advisor

Leslie Ying

Committee Members

Kevin F. King, Yi Hu, Zeyun Yu

Keywords

Channel Reduction, Compressed Sensing, GRAPPA, Kernel Method, Magnetic Resonance Imaging, Parallel MRI Reconstruction

Abstract

Magnetic resonance imaging (MRI) has revolutionized radiology in the past four decades by its ability to visualize not only the detailed anatomical structures, but also function and metabolism information. A major limitation with MRI is its low imaging speed, which makes it difficult to image the moving objects. Parallel MRI (pMRI) is an emerging technique to increase the speed of MRI. It acquires the MRI data from multiple coils simultaneously such that fast imaging can be achieved by reducing the amount of data acquired in each coil. Several methods have developed to reconstruct the original image using the reduced data from multiple coils based on their distinct spatial sensitivities. Among the existing methods, Sensitivity Encoding (SENSE) and GeneRally Autocalibrating Partially Parallel Acquisition (GRAPPA) are commercially used reconstruction methods for parallel MRI. Both methods use linear approaches for image reconstruction. GRAPPA is known to outperform SENSE because no coil sensitivities are needed in reconstruction. However, GRAPPA can only accelerate the speed by a factor of 2-3. The objective of this dissertation is to develop novel techniques to significantly improve the acceleration factor upon the existing GRAPPA methods. Motivated by the success of recent study in our group which has demonstrated the benefit of nonlinear approaches for SENSE, in this dissertation, nonlinear approaches are studied for GRAPPA. Based on the fact that GRAPPA needs a calibration step before reconstruction, nonlinear models are investigated in both calibration and reconstruction using a kernel method widely used in machine learning. In addition, compressed sensing (CS), a nonlinear optimization technique will also be incorporated for even higher accelerations. In order to reduce the computation time, a nonlinear approach is proposed to reduce the effective number of coils in reconstruction. The imaging speed is expected to improve by a factor of 4-6 using the proposed nonlinear techniques. These new techniques will find many applications in accurate brain imaging, dynamic cardiac imaging, functional imaging, and so forth.

Share

COinS