Quantification of Fat Content and Fatty Acid Composition Using Magnetic Resonance Imaging

In obesity and several other disease scenarios, the measurement of fat accumulation in various organs and tissues has become a sought-after technique in clinical diagnostics and research. Especially, quantitative and non-invasive techniques which also provide images of accumulated fat throughout the body would be valuable. In the typical hospital, magnetic resonance imaging (MRI) is the only technique available which has the potential for these types of measurements and out of techniques suggested, Water/Fat Imaging is particularly promising. Water/Fat Imaging is based on the separation of water and fat, the two main contributors to the MRI signal, with the use of the frequency separation between their signals. The field has inspired a wide range of research and is by now well established, especially for investigations of fatty liver. However, in this and several other applications there is a continued need for method development. The fat concentrations in skeletal muscle are expected to be very low. Thus, for this application, there is an increased demand for measurement precision. The results presented in this thesis indicate that fat concentrations below 1 % are possible to measure using Water/Fat Imaging. In addition, precision may be increased using a higher flip angle if a pure fat reference is used for quantification, without compromising quantification accuracy. (Papers I and II) Water/Fat Imaging has become an appreciated technique for abdominal applications in which breath-hold is necessary for acceptable image quality. The use of a bipolar acquisition scheme may be used to reduce the total scan time, but is associated with issues which are detrimental to fat quantification. Using a built-in correction approach, it is demonstrated that accurate and noise efficient fat quantification is possible using a bipolar acquisition. (Paper III) The basic ideas of Water/Fat Imaging may be extended to not only quantify the fat concentration, but also the fatty acid composition. In this thesis, a reconstruction algorithm is suggested and its accuracy is demonstrated in a wide range of fat concentrations and fatty acid compositions. In addition, a number of potential sources of bias are investigated. Out of these, accurate modeling of the individual T2 values of fat and water is especially important in fat/water mixtures, whereas some T1 weighting may be allowed with small impact on quantification accuracy. (Papers IV and V