The Hamlyn Distinguished Lecture Series
Abstract
Magnetic resonance imaging (MRI) of the heart is becoming a clinically valuable way to evaluate the heart. In order to create high quality cine images of different cardiac cycle phases with MRI, we conventionally combine imaging data acquired from multiple heart beats, in order to reconstruct images of serial frames from an “average” cardiac cycle. However, in the presence of respiratory motion or arrhythmias, there may be inconsistencies of the image data acquired from different heart beats; this can result in blurring or other artifacts that can significantly degrade the reconstructed images. We have been exploring the use of sparsity-based methods for image acquisition and reconstruction to overcome this problem. Sparsity-based or “compressed sensing” methods take advantage of the inherent correlation between pixels found in most clinical images (i.e., “sparsifiable” images), to allow the reconstruction of good quality images even from imaging data that would have been too sparsely sampled for conventional reconstruction methods; this can already be used for accelerating the acquisition of individual cardiac MRI images. We have now extended this sparsity-based imaging approach to further take advantage of the correlations found between images that reflect cardiac or respiratory motion, which introduces an additional degree of sparsifiability. By considering the heart and its associated changes with cardiac and respiratory motion as a generalized “multidimensional” object, to be reconstructed as a whole rather than using the usual individual frame-based approach, we can reconstruct good quality sets of images of the heart in different cardiac cycle phases, at different respiratory cycle phases, even in the setting of free breathing; these can be used for conventional structural or functional analysis. We can also use the multidimensional reconstructed image data to produce images of the change in the heart as a function of respiratory cycle phase, at a given cardiac cycle phase; this can provide novel physiological information, such as on RV-LV interaction associated with breathing. As a further generalization of this approach, we can consider the duration of the cardiac cycle as an additional “dimension” to be reconstructed; this can be used to reconstruct good quality cardiac images even in the setting of arrhythmia, both of the normal and of aberrant beats. Thus, while still very much a “work-in-progress”, this new approach to cardiac MRI promises to provide high quality structural images of the heart even in the setting of free breathing or arrhythmia, as well as providing new kinds of potentially useful functional information.
Biography – Professor Leon Axel
Professor Leon Axel is a Professor of Radiology, Medicine, and Physiology and Neuroscience at the NYU School of Medicine. After completing his undergraduate studies at Syracuse University, Dr. Axel pursued a doctorate in astrophysics at Princeton University. He then completed his medical education, diagnostic radiology residency and fellowship training at the University of California, San Francisco. After 20 years on the faculty of University of Pennsylvania School of Medicine, he joined the NYU School of Medicine faculty in 2001.
Professor Axel’s clinical expertise is cardiac MRI. He is active in research, focusing on translational aspects of cardiovascular MRI, seeking to develop new means of acquiring and analyzing MR images to obtain greater information about underlying physiology and pathophysiology, while applying this information to relevant clinical applications. Specifically, he is interested in developing and testing ways to study quantitative regional function and perfusion of the heart wall.