Enhancing Magnetic Resonance Imaging By Combining Parallel Imaging And Compressive Sensing

Given the broad prevalence and profound individual and public health impact of cardiovascular disease, noninvasive assessment of cardiac structure and function has long been a target for a wide range of imaging modalities. Clinical assessment of cardiac perfusion in particular plays a key role in the diagnosis and management of patients with ischemic heart disease. Myocardial perfusion imaging by magnetic resonance, in which uptake of an injected MR contrast agent into the myocardium is monitored over time, has shown great promise as an alternative to traditional and in many ways limited techniques such as SPECT. However, perfusion MR suffers in a particularly acute fashion from competing constraints of spatial and temporal resolution which underlie most cardiac imaging techniques. In order to satisfy these stringent constraints, most cardiac perfusion MR studies have acquired images at only a small number of locations through the heart, and at comparatively coarse spatial resolution.

Traditional constraints on MR imaging speed may be circumvented by use of parallel imaging techniques. Parallel MRI uses arrays of radio frequency detector coils to generate multiple image components simultaneously rather than in a traditional sequential order. Parallel MRI allows many-fold accelerations of image acquisition, and it is now used in a substantial fraction of all MR examinations worldwide, with a particularly vigorous application in the area of cardiac imaging.

Another approach to accelerated imaging which has received increasing attention in recent times is compressed sensing. Compressed sensing techniques, which have a rich history in engineering and signal processing disciplines have only recently been making their way into the MR field. Compressed sensing applies the fact that an image is typically sparse under an appropriate representation basis to design tailored image acquisitions and reconstructions which preserve image content despite a substantial degree of undersampling.

This project aims to improve MRI by combining compressed sensing and parallel imaging, therefore fortifying the quality of reconstructed images at low sampling rates. The improvement of the image quality with fewer samples enables higher resolution scanning especially in applications such as cardiac perfusion in which the measurement time is limited.