Chemical-shift encoding-based water-fat separation with multifrequency fat spectrum modeling in spin-lock MRI.

Abstract

Chemical exchange saturation transfer is used commonly to generate MRI contrast based on the chemical exchange effect. The spin-lock techniques can also be used to probe the chemical exchange and other molecular motion processes in tissues. The presence of fat can cause errors in spin-lock MRI. Signals from fat are typically suppressed based on spectral selectivity or T1 nulling approaches in spin-lock imaging. However, these methods cannot be used to suppress fat signals from multiple fat peaks. To address this problem, we report chemical-shift encoding-based water-fat separation approaches with multifrequency fat spectrum modeling. Both the conventional spin-lock and the adiabatic continuous-wave constant-amplitude spin lock (ACCSL) with multi-echo acquisitions are investigated for chemical-shift encoding-based water-fat separation in spin-lock imaging. A comparison is made of reconstructions based on 3 models: a single-peak fat spectrum model, a standard precalibrated proton density 6-peak fat spectrum model, and the self-calibrated relaxation-dependent 3-peak fat spectrum model. Comparisons were performed using Bloch simulations, phantom, and in vivo experiments at 3 T. Conventional spin-lock acquisitions cannot be used for reliable water-fat separation with a multipeak fat spectrum model. Water-fat separation based on ACCSL acquisitions achieves superior performance compared with the use of conventional spin-lock acquisitions. The best result is achieved from ACCSL acquisition with self-calibrated relaxation-dependent multipeak fat spectrum modeling. The ACCSL acquisition can be used for chemical-shift encoding-based water-fat separation with multipeak fat spectrum modeling. This approach has the potential to improve quantitative analysis using spin-lock MRI for assessing the biochemical properties of tissues.