Medical image segmentation method based on CM-MLP network

The invention discloses a medical image segmentation method based on a CM-MLP network. The method comprises the following steps: (1) constructing a medical image segmentation network; (2) adding parallel branches as extension of the network; (3) adding an MFI module and an ACRE module to learn features of segmentation edges; (4) adding a deep supervision loss control branch learning direction; according to the invention, a branch network based on deep learning is established, and a segmentation task is carried out by using a deep supervision loss constraint edge segmentation mode; the MFI module uses a windowed full-connection neural network and can process all local information in branches; the ACRE module uses a context relation encoder based on axial attention, and can better pay attention to edge information of a segmented object; the deep supervision loss is used as a part of the total loss, and the edge information extraction capability of the parallel branches is improved; according to the method, the