ATFE-Net: Axial Transformer and Feature Enhancement-based CNN for ultrasound breast mass segmentation

• Published in: Computers in biology and medicine Vol. 153; p. 106533
• Main Authors: Ma, Zhou, Qi, Yunliang, Xu, Chunbo, Zhao, Wei, Lou, Meng, Wang, Yiming, Ma, Yide
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.02.2023; Elsevier Limited
• Subjects: Axial transformer; Breast cancer; Breast mass segmentation; Convolutional neural network; Feature enhancement; Feature maps; Image Processing, Computer-Assisted; Modules; Segmentation; Self-attention; Semantics; Transformers; Ultrasonic imaging; Ultrasonography; Ultrasound; Axial transformer; Feature enhancement; Breast mass segmentation; Convolutional neural network; Self-attention
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2022.106533

Breast mass is one of the main clinical symptoms of breast cancer. Recently, many CNN-based methods for breast mass segmentation have been proposed. However, these methods have difficulties in capturing long-range dependencies, causing poor segmentation of large-scale breast masses. In this paper, we propose an axial Transformer and feature enhancement-based CNN (ATFE-Net) for ultrasound breast mass segmentation. Specially, an axial Transformer (Axial-Trans) module and a Transformer-based feature enhancement (Trans-FE) module are proposed to capture long-range dependencies. Axial-Trans module only calculates self-attention in width and height directions of input feature maps, which reduces the complexity of self-attention significantly from O(n2) to O(n). In addition, Trans-FE module can enhance feature representation by capturing dependencies between different feature layers, since deeper feature layers have richer semantic information and shallower feature layers have more detailed information. The experimental results show that our ATFE-Net achieved better performance than several state-of-the-art methods on two publicly available breast ultrasound datasets, with Dice coefficient of 82.46% for BUSI and 86.78% for UDIAT, respectively. •An Axial Transformer (Axial-Trans) module is proposed.•A Transformer-based feature enhancement (Trans-FE) module is proposed.•We develop an axial Transformer and feature enhancement-based CNN (ATFE-Net).