Multi-channel magnetic resonance imaging method based on CMA-Net

The invention discloses a multichannel magnetic resonance imaging method based on CMA-Net. The method comprises the steps of k space data preprocessing, GRAPPA reconstruction, construction and training of a CMA-Net network model, data consistency replacement and image reconstruction. The CMA-Net network model uses a plurality of multi-scale cascade attention modules to improve the global feature extraction capability of the network, and a plurality of attention gate modules are embedded into rapid connection between a coding path and a decoding path to improve the feature extraction capability of the network on local region-of-interest information; residual connection is constructed by using a GRAPPA reconstruction result, and the imaging quality is further improved by jointly utilizing the advantages of linear interpolation and nonlinear interpolation; and multi-angle residual connection is constructed, so that the gradient disappearance problem is effectively relieved, and the network performance is improve