Figure 2
From: Medical image segmentation model based on triple gate MultiLayer perceptron
(a) shows the main architecture of the TGMLP U-Net with local-global training strategy. (b) Indicates that Encoder in TGMLP U-Net uses TGMLP and Convolution. (c) indicates that Decoder in TGMLP U-Net uses \(3\times 3\) Convolution layer, Deconvolution and jump connection (the Layer Norm and residual connection are omitted in Fig. 2b).
