TransUNet: Rethinking the U-Net architecture design for medical image segmentation through the lens of transformers

Jieneng Chen, Jieru Mei, Xianhang Li, Yongyi Lu, Qihang Yu, Qingyue Wei, Xiangde Luo, Yutong Xie, Ehsan Adeli, Yan Wang, Matthew P. Lungren, Shaoting Zhang, Lei Xing, Le Lu, Alan L. Yuille, Yuyin Zhou

Abstract: Highlights•Incorporating self- and cross-attention into U-Net for medical image segmentation•Transformer decoder with coarse-to-fine attention to enhance small tumor segmentation•TransUNet enhances U-Net’s encoding/decoding, surpassing nnUNet on multiple tasks.•Our codebase supports 2D and 3D implementations to foster future exploration.