Article
Biology
Zhou Ma, Yunliang Qi, Chunbo Xu, Wei Zhao, Meng Lou, Yiming Wang, Yide Ma
Summary: This paper proposes an axial Transformer and feature enhancement-based CNN (ATFE-Net) for ultrasound breast mass segmentation. The ATFE-Net utilizes an axial Transformer module and a Transformer-based feature enhancement module to capture long-range dependencies and enhance feature representation. Experimental results demonstrate that the ATFE-Net outperforms several state-of-the-art methods on breast ultrasound datasets.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Biology
Jintao Ru, Beichen Lu, Buran Chen, Jialin Shi, Gaoxiang Chen, Meihao Wang, Zhifang Pan, Yezhi Lin, Zhihong Gao, Jiejie Zhou, Xiaoming Liu, Chen Zhang
Summary: Breast cancer is the most common cancer in women. Ultrasound and DCE-MRI are widely used for diagnosis, and we propose a segmentation network named Att-U-Node to address the challenges of current deep neural networks. The experiments show that the proposed model achieves competitive results for tumor segmentation and mitigates the common problems of deep neural networks.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Automation & Control Systems
Gongping Chen, Yu Dai, Jianxun Zhang
Summary: In this paper, a novel refinement residual convolutional network is developed to accurately segment breast tumors from ultrasound images. The network combines deep learning methods, including SegNet with deep supervision module, missed detection residual network, and false detection residual network. Experimental results demonstrate that the proposed method achieves the best segmentation results, indicating its superior adaptability for breast tumors segmentation.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Mathematical & Computational Biology
Rongrong Bi, Chunlei Ji, Zhipeng Yang, Meixia Qiao, Peiqing Lv, Haiying Wang
Summary: This study proposes a new end-to-end network called ResCEAttUnet to improve the segmentation accuracy of liver tumors from CT images. The experimental results demonstrate the significant improvement in segmentation accuracy compared to state-of-the-art methods, indicating the promising potential of the proposed method.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2022)
Article
Biochemical Research Methods
Minghao Xue, Menghao Zhang, Shuying Li, Yun Zou, Quing Zhu
Summary: In this study, an automated ultrasound (US)-assisted diffuse optical tomography (DOT) pipeline was developed for fast and accurate diagnosis. The pipeline includes automated DOT pre-processing, imaging reconstruction, and final diagnosis based on US imaging features and DOT measurements. This approach significantly reduces the processing time while maintaining comparable classification results with manual processing.
BIOMEDICAL OPTICS EXPRESS
(2023)
Article
Engineering, Biomedical
Tao Jiang, Wenyu Xing, Ming Yu, Dean Ta
Summary: This article presents a novel U-shaped segmentation model based on a hybrid CNN-transformer structure for ultrasound image segmentation. The results demonstrate that our method outperforms state-of-the-art segmentation methods on four public ultrasound datasets, achieving excellent results in tasks such as breast lesion segmentation, thyroid lesion segmentation, and left ventricle segmentation.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Wanli Ding, Heye Zhang, Shuxin Zhuang, Zhemin Zhuang, Zhifan Gao
Summary: This paper proposes a multi-view stereoscopic attention network (MVSA-Net) to address the challenges of locating and classifying lesions in automated breast ultrasound (ABUS) images. MVSA-Net crops the three-dimensional lesion regions and classifies them through the stereoscopic features extracted from the Transformer network. The results show that MVSA-Net achieves high accuracy and AUC, indicating its potential to assist diagnosticians in reducing the interpretation time of ABUS images.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Mathematical & Computational Biology
Yuqing Zhang, Yutong Han, Jianxin Zhang
Summary: This paper proposes a novel mixed attention U-Net model, called MAU-Net, for MRI brain tumor segmentation. The model combines spatial-channel attention and self-attention, achieving significant improvements in local and global information modeling.
MATHEMATICAL BIOSCIENCES AND ENGINEERING
(2023)
Article
Engineering, Biomedical
Yuchao Lyu, Yinghao Xu, Xi Jiang, Jianing Liu, Xiaoyan Zhao, Xijun Zhu
Summary: This study proposes an improved Pyramid Attention Network (AMS-PAN) for accurate segmentation of breast ultrasound images. The model combines attention mechanism, multi-scale feature extraction, and spatial/channel attention module to achieve good segmentation performance. It has proven to be efficient in assisting physicians in breast tumor ultrasound detection tasks and guiding subsequent diagnosis and treatment services for patients.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Vivek Kumar Singh, Elham Yousef Kalafi, Shuhang Wang, Alex Benjamin, Mercy Asideu, Viksit Kumar, Anthony E. Samir
Summary: Medical image segmentation is crucial for diagnosing and staging diseases. We propose a robust and lightweight deep learning real-time segmentation network called MISegNet, which outperforms state-of-the-art methods on multiple datasets, demonstrating its versatility and effectiveness.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Biomedical
Zheng Huang, Yiwen Zhao, Yunhui Liu, Guoli Song
Summary: The study introduces a group cross-channel attention residual UNet (GCAUNet) for brain tumor segmentation, which significantly improves performance by incorporating detail recovering paths and cross-channel attention modules.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Engineering, Biomedical
Huaikun Zhang, Jing Lian, Zetong Yi, Ruichao Wu, Xiangyu Lu, Pei Ma, Yide Ma
Summary: In this paper, the authors propose a hybrid CNN-transformer framework called HAU-Net for breast lesion segmentation. By combining the long-distance dependence of transformers and the local detail representation of CNNs, this framework achieves better performance in segmenting challenging breast ultrasound images.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2024)
Article
Biology
Jianning Chi, Zelan Li, Zhiyi Sun, Xiaosheng Yu, Huan Wang
Summary: This paper proposes a novel hybrid transformer UNet (H-TUNet) for thyroid gland segmentation in ultrasound sequences. It integrates and refines low-level features from different encoding layers using a designed multi-scale cross-attention transformer (MSCAT) module. It also strengthens contextual features from successive frames using a 3D self-attention transformer (SAT) module. Experimental results on TSUD and TG3k datasets demonstrate the superiority of the proposed method in thyroid gland segmentation.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Computer Science, Information Systems
Mingjia Wang, YuCui Chen, Baozhu Qi
Summary: This paper proposes a novel automatic segmentation method for rectal tumors based on deep learning methods. The method utilizes a residual UNet network model that combines spatial attention and channel attention. Experimental results show that this method can effectively segment the rectal tumor area and achieve better evaluation indicators compared to traditional methods.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Artificial Intelligence
Lina Cai, Qingkai Li, Junhua Zhang, Zhenghua Zhang, Rui Yang, Lun Zhang
Summary: The study proposes a hybrid architecture based on Transformer and U-Net for ultrasound image segmentation. By using joint loss for optimization, the method achieves good performance in terms of accuracy and generalization.
PEERJ COMPUTER SCIENCE
(2023)