The overview of Residual Non-Local Fourier Network (RNLFNet) and its data flow are illustrated in Fig. 1. Given the input degraded MR image Iu, the reconstructed MR image IR can be obtained as IR=HRNLFNetIu, where HRNLFNet denotes the model of the proposed RNLFNet. The proposed model maims to learn the degradation components of the degraded MR images. In this work, our RNLFNet is the data-driven model, which learns an end-to-end mapping between zero-filled and fully sampled MR images. Here, we