An iterative multi-stage MR image correction method

Nowadays multiple receiver coil systems are often used in magnetic resonance imaging. The sensitivity profiles of these coils are usually unknown and could be quite different so that the sum-of-squares (SoS) reconstruction algorithm can yield less desirable performance. In the previous research, a polynomial is used to iteratively fit the coil sensitivity profiles and the reconstruction performance is improved. We extend that algorithm by considering multi-stage reconstruction. More specifically, a lower order polynomial is used at the first stage when the noise is strong. After that, the noise has been efficiently attenuated and a higher order polynomial is used to better fit the coil sensitivity profiles. Experimental results confirm the efficiency of this multi-stage algorithm.