An Improved Algorithm of Motion Artifacts Correction for MRI

To further suppress the motion artifacts and improve the MRI image quality, this paper presents an improved PROPELLER algorithm using frequency domain and image domain information to estimate motion parameter based on wavelet transformation. Different from the conventional PROPELLER algorithm, in the improved algorithm, the temporary image reconstruction of the k-space data strip is firstly obtained, and then the wavelet decomposition is performed, finally, motion parameters are searched in the case of a multi-resolution, thus the motion artifacts correction is reached. The essence of the improved algorithm is using the normalized mutual information (NMI) as the similarity measure to estimate the motion parameters. After motion compensation, the image was obtained through gridding reconstruction. In the process of motion parameter estimation, not only taking into account the amplitude of the frequency domain data, the change of phase is also considered, so the accuracy of rotation motion parameter estimation could be improved. To demonstrate the validity of the improved algorithm, the simulation experiments were performed to the conventional PROPELLER algorithm and the improved algorithm. The Shepp-Logan (SL) template image and phantom image experiments show that the improved method can reduce the motion artifacts more effectively by comparing it with the conventional PROPELLER method.