The USC Andrew and Erna Viterbi School of Engineering USC Signal and Image Processing Institute USC Ming Hsieh Department of Electrical Engineering University of Southern California

Technical Report USC-SIPI-397

“Dynamic Cardiovascular Magnetic Resonance Imaging for Improved Assessment of Ischemic Heart Disease”

by Taehoon Shin

December 2009

Magnetic resonance imaging (MRI) is one of the most powerful non-invasive imaging modalities whose strengths include flexible tissue contrast, sensitivity to flow, and the lack of any harmful radiation. Ischemic heart disease (IHD) is a leading cause of death in the western world, and as such, MR approaches for the assessment of IHD have gained considerable attention. This dissertation will describe two improvements in cardiac MRI methodology that are aimed at improving the assessment of IHD: (i) Real-time cardiac imaging and (ii) First-pass myocardial perfusion imaging.The first method, real-time cardiac MRI continuously captures the beating heart without the need for cardiac gating and breathhold, and is a promising approach for the assessment of ventricular function. Efficient k-space sampling trajectories (e.g., spiral or echo-planar) are typically used to reduce acquisition time, but further improvement in spatio-temporal resolution is desirable. A novel reconstruction method was developed for accelerating real-time spiral imaging, based on an algebraic reconstruction framework that can be applied to undersampled data. The method utilizes knowledge of the aliasing pattern caused by spiral undersampling, and the fact that temporal bandwidth of dynamic cardiac images is a function of spatial position.The second method, first-pass myocardial perfusion imaging (MPI) is a tool for early detection of IHD by evaluating blood flow to the myocardium. The most widely used protocol for MR MPI involves a 2D multi-slice acquisition, which has demonstrated high sensitivity and specificity, but suffers from incomplete spatial coverage. We developed a three-dimensional MR MPI approach that allows contiguous and complete coverage of the left ventricle. We demonstrate its accuracy in a cardiac phantom, and its feasibility in several healthy volunteers. We also compare the performance of diastolic acquisition, which has a longer quiescent period and enables higher spatial resolution, versus systolic acquisition, which is less sensitive to heart-rate variation or arrhythmia. Achieving both complete coverage and high spatial resolution is challenging in 3D MPI. The last part of this dissertation focuses the use of regularized reconstruction, and how it may be used to improve spatial resolution.

To download the report in PDF format click here: USC-SIPI-397.pdf (7.0Mb)