Fast methods for MR spectroscopic imaging

Magnetic Resonance Spectroscopic imaging is an emerging, non-invasive and non-destructive technique for the evaluation of metabolism in vivo. Limited data sampling is due to difficulties inherent in this technique. The SLIM and GSLIM methods have been proposed for the reconstruction of spectral functions from few phase-encoded data. However, the practical application of these methods is limited. SLIM recovers efficiently the spectral function only if the imaged object is correctly divided into homogeneous compartments, while GSLIM has a high computational complexity. In this work, we propose two new methods which are able to model inhomogeneity into the imaged object and have a low computational complexity.