Contourlet Based MR Image Reconstruction via Reweighted L1-Minimization

How to improve MR imaging speed by using compressive sensing (CS) theory, which reduce imaging time by sampling only a small bit of K-space data, has become a research focus of MR reconstruction. The measurement required in CS theory relies on the sparsity of sparse transform and the objective function of reconstruction. In this paper, we introduce contourlet transform into MR reconstruction based on CS, which achieves a sparser representation of image in contrast with orthogonal transforms such as wavelet. In addition, a reweighted L1-minimization scheme is introduced as the objective function instead of the traditional unweighted L1-minimization to further improve reconstruction performance. The experimental results prove the effectiveness of the proposed approach.