Improved SENSE imaging using accurate coil sensitivity maps generated by a global magnitude-phase fitting method.

Abstract

To develop a novel coil sensitivity processing technique that is able to reduce or eliminate aliasing artifacts and noise amplification in image-domain parallel imaging (i.e., SENSE). Initial coil sensitivity maps were derived from the center k-space lines by a common self-calibration method. Then continuous trigonometric functions were used to fit both magnitude and phase maps of the self-calibrated coil sensitivity profile globally. Further, the global fitted coil sensitivity maps were adopted for SENSE reconstruction. Numerical simulations, as well as experiments on phantoms and human subjects were performed to evaluate and compare the effectiveness of this global magnitude-phase fitting approach with traditional local fitting methods. Both simulation and experimental results demonstrated that the proposed novel global fitting method was able to obtain accurate coil sensitivity profiles without Gibbs oscillations. The resultant SENSE images were improved substantially in terms of aliasing imaging artifacts. A global magnitude-phase fitting method for better estimation of accurate coil sensitivity maps was developed, and it was successfully used in producing high-quality parallel images. Magn Reson Med 74:217-224, 2015. © 2014 Wiley Periodicals, Inc.