In dynamic magnetic resonance (MR) imaging, low-rank plus sparse (L+S) decomposition, or robust principal component analysis (PCA), has achieved stunning performance. However, the selection of the parameters of L+S is empirical, and the acceleration rate is limited, which are common failings of iterative compressed sensing MR imaging (CS-MRI) reconstruction methods. Many deep learning approaches have been proposed to address these issues, but few of them use a low-rank prior. In this paper, a model-based low-rank plus sparse network, dubbed L+S-Net, is proposed for dynamic MR reconstruction. In particular, we use an alternating linearized minimization method to solve the optimization problem with low-rank and sparse regularization. Learned soft singular value thresholding is introduced to ensure the clear separation of the L component and S component. Then, the iterative steps are unrolled into a network in which the regularization parameters are learnable. We prove that the proposed L+S-Net achieves global convergence under two standard assumptions. Experiments on retrospective and prospective cardiac cine datasets show that the proposed model outperforms state-of-the-art CS and existing deep learning methods and has great potential for extremely high acceleration factors (up to 24x).
翻译:在动态磁共振成像(MR)中,低声加稀少(L+S)分解(L+S)分解(L+S)的动态磁共振成像(MR)成像(MR)成象(MR)的动态磁共振(MR)成象(MR)成象(MR(MR)成象)的重建方法中,选择L+S的参数是经验性的,加速率是有限的,这是反复压缩感(CS-MRI)M(CS-MRI)成象(CS-MRI)成象(CS-MRI)成象(CS-MRI)成象(CS-MRI)成象(C-MRI)成象的常见失败。许多深层次的学习方法是为了解决这些问题,但很少使用以前低级的方法。在本文件中,为动态的MRMRM重建提议了一个以基于模式、低点加零的网络(Dbbbed L+L+S-L+S网(L+S-S-S-S-Net)网(L+S-L+S-S-S-Net网)的模型,称为DMRMRMRV的模型实验,称为D-未来的热心心肾-未来的试验显示显示显示显示,特别是实验显示,特别的显示,特别是我们追溯和未来的最深的实验显示,特别是前的模型实验显示,特别是深点的实验显示,特别,特别是我们用的极的极的模型已显示的实验显示,用最深深点的实验显示,用模型的实验显示,以低的模型已超过模型的模型的模型的模型的模型的深点的模型的模型已超过24国和深点,和深室-S-S-S-S-S-S-S-S-S-C室-S-C室-C室-C室-C室-C室-S-S-S-S-S-C室-C室-S-HIS-S-S-C室-S-S-S-C室-C室-S-S-S-S-S-S-S-S-C室-S-C室-C室-C室-C室-C室-C的