Fig. 4
From: Enhanced glioma semantic segmentation using U-net and pre-trained backbone U-net architectures
Detailed architecture of the Inception-U-Net for glioma semantic segmentation. The architecture is depicted in two main sections: the encoder, which utilizes Inception blocks for multi-scale feature extraction, and the decoder, which employs up-convolution layers for precise localization. Skip connections between corresponding layers in the encoder and decoder facilitate the integration of contextual information at various resolutions. The final output layer uses a sigmoid activation function to produce the segmented image.
