A new technique for homonuclear broadband-decoupled magnetic resonance chemical shift imaging.

Abstract

A novel NMR chemical shift imaging technique which provides an economical means of gathering spectrally encoded data, inherently J-decoupled, for image formation is described. This technique, called SLIM-CSI, employs a constant-time (or generalized) Dixon sequence, with a set of optimized time-displacements that encode the chemical shift information. The optimization criterion is based on the minimization of the condition number of the kernel matrix which is functionally dependent upon a pre-determined chemical shift spectrum of the cross-sectional area of the object. Because the number of data lines required is small in comparison to the conventional FT technique, this technique can produce chemical shift-segregated images with relatively high resolution and greatly reduced data acquisition time. SLIM-CSI is particularly advantageous for systems that are anatomically or histologically complex but spectroscopically simple because the spectral information in this technique is phase- rather than frequency-encoded. (The SLIM-CSI technique becomes the well-known Dixon technique in the extreme case of a two-component system.) The optimization of time displacements for chemical shift encoding significantly improves the quality of spectral separation. The technique contains in situ static field inhomogeneity correction and is inherently broadband J-decoupled. Unraveling or reconstruction of spectral component images from the chemical shift-encoded image data is formulated as a least squares problem solved by singular value decomposition. The optimization is performed by the downhill simplex algorithm. Three types of experiments demonstrate the feasibility and possible utility in biomedical applications. A simulated experiment illustrates the fundamental soundness of the technique, the significance of time-displacement optimization and the effect of field correction, and provides insights on how the various parameter misadjustments or hardware imperfection may affect image separation. Phantom experiments provide empirical verification of the simulated results and serve as a prelude to intact animal experiments. In vivo experiments using rat brains demonstrate possible applications of this technique in a biological setting.