A Fast dynamic mr imaging method using compressed sensing and multiple transforms

Abstract

A dynamic magnetic resonance imaging (MRI) such as cardiac MRI or functional MRI (fMRI) requires a significant reduction of the data acquisition time in order to capture the cardiac motion or blood-oxygenation level dependent (BOLD) response in a short time. The conventional approaches to the dynamic imaging include parallel imaging, generalized series (GS), higher-order generalized series (HGS) and their combinations. Recently, compressed sensing (CS) is an emerged technique for reconstructing signals from sampled data under the Nyquist rate, which uses the assumption of sparse representation of desired signal in a sparsifying transform domain. Since the dynamic imaging has the redundant data repetitively, the undersampled data is enough to reconstruct images using fully sampled reference data. In this paper, we propose a novel reconstruction method that estimates the undersampled dynamic MR image which is sparsely represented by Karhunen-Loeve transform (KLT) and wavelet transform by using the fully sampled reference image. In order to evaluate the proposed method, we compared the proposed method with conventional method in computer simulation and experiments. In case of high reduction factor, the simulation and experimental results show that the proposed method can reconstruct dynamic signal better than those of conventional method.