Sparsity-Based MRI Reconstruction of Physiologic Dimensions

New York University School Of Medicine, New York NY

Investigators

Linked publications & trials

Abstract

? DESCRIPTION (provided by applicant): While magnetic resonance imaging (MRI) is clinically very valuable, current imaging methods are subject to blurring and artifacts in the presence of physiologic (e.g., respiratory and cardiac) motion, as well as of arrhythmias, thus limiting the practical application of MRI in many patients. The currently used MRI methods are also limited in their ability to study the effects of free breathing and arrhythmias on the heart. The proposed research will further develop and evaluate a new approach to imaging in the presence of physiologic motion, which parameterizes such motions with a variable that is treated as an additional dimension to be reconstructed. This would not be practically feasible with conventional methods, due to the additional associated data acquisition that would be required. However, with the use of sparsity-based image reconstruction methods, the high degree of correlation of the images along these additional dimensions permits good quality image reconstructions, even with heavily undersampled imaging data. We already have made successful initial implementations of this new method for 2D cine imaging and 3D MR angiography. In the proposed research, we will further improve these initial implementations, and we will extend them to include implementations of our methods for other MRI sequences, particularly fully 3D cine data acquisitions. We will evaluate the image quality achievable with these new methods in the presence of free breathing and arrhythmias, as compared with conventional clinical imaging methods, using both numerical phantom simulations and clinical cardiac function analysis in pediatric patients to test the performance. We will also evaluate the potential for extracting new kinds of functional information from these multidimensional image sets, including the effects of free breathing and arrhythmias on the heart, using analysis tools that we will be developing. If this research is successful, these new methods will provide significantly improved MR image quality in the presence of free breathing and arrhythmias, as well as providing potentially valuable new kinds of information on the function of the heart. They may also be able to be used for performing MRI in the presence of exercise, which could be useful for both cardiovascular and musculoskeletal applications, as well as in combination with other kinds of imaging, such as with integrated PET/MRI systems. This research should thus further increase the clinical utility of MR imaging for many patients.