Abstract: Current methods in medical image semantic segmentation often rely on simple skip connections within U-shaped network structures. These approaches fail to bridge the semantic gap between the ...