Abstract: Medical image segmentation can assist doctors to quickly and accurately identify organs and lesions in medical images, which is of great value in improving the efficiency of clinical diagnosis. U-Net combined with Transformer is the mainstream method in the field of medical image segmentation. However, Transformer has a weak ability to extract local information, and the U-Net structure will lose detailed location information during upsampling and subsampling. To address the above problems, this paper proposes a TransUnet medical image segmentation network with enhanced position information, PETransUnet. The network first uses the Positional Efficient Attention Block (PEA) to enhance the position information of features; secondly, the Dual Attention Bridge Block (DAB) is used to make up for the semantic gap between the features in the encoding stage and the decoding stage; finally, the Cross-channel Attention Fusion Block (CCAF) is used to reduce the position information lost during upsampling. The proposed method was validated on the publicly available Synapse dataset, achieving Dice coefficients of 82.92% and HD95 coefficients of 18.87%. On the ACDC dataset, a Dice coefficient of 90.73% was attained. On the LIST17 dataset, the Dice coefficients for liver and liver tumor segmentation were 94.85% and 74.47%, respectively. Comparative analysis with recent algorithms shows higher segmentation accuracy.

Key words: medical image segmentation, Transformer, feature fusion, position encoding

摘要： 医学图像分割能够辅助医生快速准确的识别医学图像中的器官和病变部位，对提高临床诊断的效率有重要的价值。结合Transformer的U-Net是当前医学图像分割领域的主流方法，但是Transformer对于局部信息的提取能力较弱，并且U-Net结构在上采样和下采样过程中会损失细节位置信息。针对以上问题，本文提出一种位置信息增强的TransUnet医学图像分割网络PETransUnet。该网络首先使用位置高效注意力模块(Positional Efficient Attention Block，PEA)对特征的位置信息进行增强；其次使用双注意力桥模块(Dual Attention Bridge Block，DAB)，弥补编码阶段和解码阶段特征之间的语义差距；最后使用跨通道注意力融合模块(Cross-channel Attention Fusion Block，CCAF)减少上采样时丢失的位置信息。提出的方法在公开数据集Synapse上进行验证，Dice系数和HD95系数分别达到82.92%和18.87%；在公开数据集ACDC上进行验证，Dice系数达到90.73%；在公开数据集LIST17上进行验证，肝脏和肝肿瘤Dice系数分别达到94.85%和74.47%。与近期多种算法进行比较，具有更高的分割精度。

关键词: 医学图像分割, Transformer, 特征融合, 位置编码