基于张量低秩加稀疏模型的MRI重建方法

MRI Reconstruction Based on Tensor Low-rank and Sparse Model

传统的基于压缩感知的磁共振(MRI)重建方法处理动态MRI数据时,需要将其向量化,但这种操作会在一定程度破坏数据内部的结构信息,从而有可能会降低重建质量.本文直接面向张量数据,建立了动态MRI重建的张量稀疏加低秩模型框架,并设计了一个基于交替方向乘子法(ADMM)的求解算法.实验结果表明,本文提出的模型和算法是有效的.

When traditional magnetic resonance(MRI)reconstruction methods based on compressed sensing process dynamic MRI data,they need to be vectorized.But this operation may destroy the inner data structures and consequently results in worse performance.In this paper,A tensor low-rank plus sparse model for dynamic MRI reconstruction is proposed,and a ADMM based algorithm is designed for solving this model.The experimental results show that compared to some of state-of-the-art methods,the proposed approach as well as its improved performance is effective.