|
|
|
“Accelerated Chemical Shift Encoded Water-Fat Imaging”
by Samir D. Sharma
May 2012
Water-fat imaging techniques are commonly used in the research and clinical setting to separate and/or quantitate the two species. Within this class of techniques, chemical shift encoded methods have recently gained significant interest because of their insensitivity to magnetic field inhomogeneity. The benefits of chemical shift encoding come at a cost of a long scan time because data must be acquired at multiple echo times. This cost increases the likelihood of image artifacts that result from patient motion during the scan while also reducing the overall level of patient comfort. The long scan time becomes especially problematic when imaging non-stationary regions of the body such as the abdomen. In these cases, a compromise with other parameters must be made. Most often, a sacrifice of the spatial resolution and/or volume coverage is required to avoid motion-related image artifacts. The objective of this dissertation is to present a new method for chemical shift encoded water-fat separation and quantitation from an accelerated data acquisition. In this context, accelerating the acquisition implies that a fewer number f measurements are acquired as compared to the requirements from the Nyquist Sampling Theorem. The resulting image recovery is an ill-posed inverse problem that requires the incorporation of additional knowledge to find the desired solution. A combined parallel imaging and compressed sensing approach for accelerated water-fat separation and quantitation is introduced. To enable this work, I developed a novel method for B0 field map estimation as an alternative to traditional region-growing and region-merging schemes. Uniform water-fat separation is demonstrated in a variety of anatomies from both retrospectively and prospectively under sampled acquisitions. In addition, accurate quantitation is presented in water-fat phantoms as well as in liver datasets acquired from seven subjects.