Fast dynamic magnetic resonance imaging using sparse recovery methods and novel signal encoding formulations
MetadataShow full item record
Magnetic resonance imaging (MRI) is a non-invasive imaging modality that provides excellent soft tissue contrast without using ionizing radiations. These qualities/properties make MRI the preferred imaging modality for critical organs like heart and brain. Over the past decade, the advancement in hardware and image reconstruction algorithms has led to substantial improvements in MRI in terms of imaging speeds, quality and reliability. However, MRI speeds need to be further improved while retaining/maintaining the image quality given that the emerging medical diagnostic procedures are increasingly relying on detailed characterization of physiological functions that evolve on time scales too fast to be captured using conventional MRI methods. This dissertation starts with presenting a sparse signal recovery based fast MRI method. This method synergistically combines a data redundancy scheme for high frequency details with a novel and physically realizable MR signal encoding formulation. The new signal encoding formulation uses clinically deployed tagging radio frequency pulses to mix information in the spatial frequency domain prior to acquisition. Thus, the new formulation leads to a more uniform coverage of spatial frequency information even at high accelerations. The synergistic combination of image-detail redundancy encoding with tagging based signal encoding allows recovery of edges and fine structures with unprecedented quality. Next, this dissertation evaluates the use of fast spiral trajectories for high spatial resolution functional imaging of human superior colliculus. Gradient efficient and motion-robust spiral trajectories are used to keep fMRI scan durations short. . However, high resolution imaging of human subcortical structures using these trajectories is limited due to the weak functional responses of SC structures and also low signal-to-noise ratio associated with small voxels. To improve the functional sensitivity of spiral trajectories, dual echo variants are used. Combination of two echoes of the dual-echo variants reduces noise and thereby improves the functional sensitivity of high resolution fMRI. Lastly, this dissertation presents a novel formulation for fast dynamic MRI which combines the generic linear dynamical system model with sparse recovery techniques. Specifically, the formulation uses a known prior spatio-temporal model to predict the underlying image and uses sparse recovery techniques to recover the residual image. The spatio-temporal evolution model inherently encodes for coupled data redundancies in the spatial- and temporal-dimensions. Also, the generalizability of the formulation in choosing the evolution model allows it to be applicable to various physiological functions.