Department of Radiology
NYU School of Medicine
March 11, 2014
Sparsity-based Cardiovascular Magnetic Resonance Imaging (MRI) of Physiologic Dimensions
Magnetic resonance imaging (MRI) has become a valuable tool to image many parts of the body, including the cardiovascular system. In cardiovascular MRI (CMR), we typically combine data from multiple cardiac cycles to reconstruct cardiac-synchronized “cine” views of the cardiac motion. However, inconsistencies in the imaging data, e.g., from respiratory motion or arrhythmias, can lead to degraded CMR image quality. We have been using sparsity-based compressed sensing approaches to CMR image reconstruction, in order to both speed up the imaging and make it more robust to such degradation. In particular, we can treat the cardiac and respiratory cycle phases as being additional effective “physiologic” dimensions to be reconstructed, and cardiac cycle perturbations due to common arrhythmias can be treated as another sort of dimension, which improves the data redundancy or sparsity of the data. The high degree of image correlation found along these dimensions allows us to efficiently create high quality cardiac- and respiratory-synchronized images with compressed sensing-based approaches, even in the presence of free breathing and arrhythmias, which would be difficult with conventional imaging approaches. Furthermore, these images can reveal novel kinds of physiologic data, such as the interaction between the right and the left sides of the heart during free breathing.
This work has been done in collaboration with Li Feng and Ricardo Otazo.