UNet 3+: A Full-Scale Connected UNet for Medical Image Segmentation

Huang, Huimin; Lin, Lanfen; Tong, Ruofeng; Hu, Hongjie; Zhang, Qiaowei; Iwamoto, Yutaro; Han, Xian-Hua; Chen, Yen‐Wei; Wu, Jian

doi:10.1109/icassp40776.2020.9053405

Cited by 1,617 publications

(750 citation statements)

References 7 publications

Supporting

Mentioning

747

Contrasting

Unclassified

Order By: Relevance

“…However, neither the raw U-Net nor the U-Net++ extracts fully the multiscale features within the network. Thus, full-scale skip connections are designed in U-Net 3+ [13] to alleviate this limitation, while leading to huge computational complexity. Besides, full-scale skip connections assume that all channels of the feature maps generated by different layers share equal weights.…”

Section: Introductionmentioning

confidence: 99%

“…Based on the insights above, we design a multi-scale connected and asymmetric-convolution-based U-Net (MACU-Net) with asymmetric convolution blocks. To test the effectiveness of MACU-Net, we compare the performance of the proposed method with U-Net [11], FGC [19], U-Net++ [12], U-NetPPL [20], WRAU-Net [21], CE-Net [22] and U-Net 3+ [13]. The major contributions of this Letter are listed as: 1) We design multi-scale skip connections with channel attention blocks to make use of multi-scale features and realign channel-wise features.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

MACU-Net for Semantic Segmentation of Fine-Resolution Remotely Sensed Images

Duan

Zheng

et al. 2022

IEEE Geosci. Remote Sensing Lett.

View full text Add to dashboard Cite

Semantic segmentation of remotely sensed images plays an important role in land resource management, yield estimation, and economic assessment. U-Net, a deep encoderdecoder architecture, has been used frequently for image segmentation with high accuracy. In this Letter, we incorporate multi-scale features generated by different layers of U-Net and design a multi-scale skip connected and asymmetric-convolutionbased U-Net (MACU-Net), for segmentation using fine-resolution remotely sensed images. Our design has the following advantages: (1) The multi-scale skip connections combine and realign semantic features contained in both low-level and high-level feature maps; (2) the asymmetric convolution block strengthens the feature representation and feature extraction capability of a standard convolution layer. Experiments conducted on two remotely sensed datasets captured by different satellite sensors demonstrate that the proposed MACU-Net transcends the U-Net, U-NetPPL, U-Net 3+, amongst other benchmark approaches.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

MACU-Net for Semantic Segmentation of Fine-Resolution Remotely Sensed Images

Duan

Zheng

et al. 2022

IEEE Geosci. Remote Sensing Lett.

View full text Add to dashboard Cite

show abstract

“…Due to the dominance of Unet, recent studies have focused on further improving the structure of Unet for application on medical image segmentation. The approaches of these studies were to change the internal structure of the nodes in the encoder and decoder blocks [ 29 , 30 , 31 , 32 ] or change the connection between the blocks [ 33 , 34 ]. Other approaches were to change the skip connection of the conventional Unet architecture [ 9 , 35 , 36 , 37 ].…”

Section: Introductionmentioning

confidence: 99%

“…Improving skip connection performance will lead to an increase in the efficiency of the entire model. Huang et al [ 33 ] used the dense skip connections to combine all of the features from the encoder node with the features from the decoder node. The full-scale aggregated feature maps are learned by deep supervision.…”

Section: Introductionmentioning

confidence: 99%

TMD-Unet: Triple-Unet with Multi-Scale Input Features and Dense Skip Connection for Medical Image Segmentation

et al. 2021

View full text Add to dashboard Cite

Deep learning is one of the most effective approaches to medical image processing applications. Network models are being studied more and more for medical image segmentation challenges. The encoder–decoder structure is achieving great success, in particular the Unet architecture, which is used as a baseline architecture for the medical image segmentation networks. Traditional Unet and Unet-based networks still have a limitation that is not able to fully exploit the output features of the convolutional units in the node. In this study, we proposed a new network model named TMD-Unet, which had three main enhancements in comparison with Unet: (1) modifying the interconnection of the network node, (2) using dilated convolution instead of the standard convolution, and (3) integrating the multi-scale input features on the input side of the model and applying a dense skip connection instead of a regular skip connection. Our experiments were performed on seven datasets, including many different medical image modalities such as colonoscopy, electron microscopy (EM), dermoscopy, computed tomography (CT), and magnetic resonance imaging (MRI). The segmentation applications implemented in the paper include EM, nuclei, polyp, skin lesion, left atrium, spleen, and liver segmentation. The dice score of our proposed models achieved 96.43% for liver segmentation, 95.51% for spleen segmentation, 92.65% for polyp segmentation, 94.11% for EM segmentation, 92.49% for nuclei segmentation, 91.81% for left atrium segmentation, and 87.27% for skin lesion segmentation. The experimental results showed that the proposed model was superior to the popular models for all seven applications, which demonstrates the high generality of the proposed model.

show abstract

“…They evaluated their method with six different datasets of medical images. Huang et al [36] also proposed the same skip connections, but combined the lowfeatures with the high-features from feature maps in different scales. They used deep supervision to learn the full-scale aggregated feature maps.…”

Section: Introductionmentioning

confidence: 99%