Wei Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Wei Chen has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
Yunfeng Ni
School of Communication and Information Engineering, Xi’an University of Science and Technology, Xi’an, China
3 shared publicationsBei Dang
School of Communication and Information Engineering, Xi’an University of Science and Technology, Xi’an, China
2 shared publicationsJie Qi
Orthopedic Department, Shaanxi Provincial People’s Hospital, Xi’an, China
2 shared publications- Jie Qi
Orthopedic Department, Shaanxi Provincial Peoples Hospital, Xian, China
2 shared publications - Mengjing Zhu
Xi’an University of Science and Technology, School of Communication and Information Engineering, Xi’an, China
2 shared publications - Pei Wang
School of Communication and Information Engineering, Xian University of Science and Technology, Shaanxi, China
2 shared publications
Latest Publications
- 26 October 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (6) , 206
YOLO-DALB: a real-time fabric defect detection algorithm based on multi-module collaborative optimization
- 26 October 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (6) , 206
YOLO-DALB: a real-time fabric defect detection algorithm based on multi-module collaborative optimization
- 14 October 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- p. 1-19
CAS-UNet: A Context-Adaptive Encoder-Decoder Network with Boundary-Distance Auxiliary Supervision for Histopathological Nuclei Segmentation
Histopathological nuclei segmentation is essential for quantitative pathological analysis but remains challenging due to staining variation, heterogeneous tissue backgrounds, diverse nuclear morphology, ambiguous boundaries, and the limited batch sizes commonly used for high-resolution image training. This study proposes CAS-UNet, a context-adaptive encoder-decoder framework with boundary-distance auxiliary supervision for binary nuclei-region segmentation. The proposed network integrates a Pathology-Adaptive Context Pyramid for adaptive multi-scale contextual aggregation, a Normalized Skip Selection Module for semantic-guided skip-feature filtering, a group-normalized residual decoder for feature reconstruction under small-batch settings, and Boundary-Distance Auxiliary Supervision for structural constraint learning. CAS-UNet was evaluated on five public histopathological nuclei datasets, including CoNIC, CPM17, MoNuSeg, NuInsSeg, and PanNuke. In the main benchmark comparison, CAS-UNet achieved the highest Dice and IoU values among the evaluated segmentation methods across the five datasets. Supplementary controlled experiments with five paired random seeds further demonstrated consistent improvements over U-Net and DeepLabV3+ under identical data partitions and training configurations. Additional experiments showed improved performance compared with traditional image-processing pipelines and more consistent segmentation results with Group Normalization than Batch Normalization under different batch sizes. Multi-seed component ablation experiments further characterized the contributions of PACP, NSSM, GN-based decoding, and BDAS. The proposed CAS-UNet provides an accurate and reliable framework for binary nuclei-region segmentation. Future work will focus on instance separation, stain-robust domain adaptation, independent clinical validation, and whole-slide image evaluation to further explore its potential for downstream pathological analysis.
- 14 October 2026
- Published by Springer Nature in Journal of Imaging Informatics in Medicine
- p. 1-19
CAS-UNet: A Context-Adaptive Encoder-Decoder Network with Boundary-Distance Auxiliary Supervision for Histopathological Nuclei Segmentation
Histopathological nuclei segmentation is essential for quantitative pathological analysis but remains challenging due to staining variation, heterogeneous tissue backgrounds, diverse nuclear morphology, ambiguous boundaries, and the limited batch sizes commonly used for high-resolution image training. This study proposes CAS-UNet, a context-adaptive encoder-decoder framework with boundary-distance auxiliary supervision for binary nuclei-region segmentation. The proposed network integrates a Pathology-Adaptive Context Pyramid for adaptive multi-scale contextual aggregation, a Normalized Skip Selection Module for semantic-guided skip-feature filtering, a group-normalized residual decoder for feature reconstruction under small-batch settings, and Boundary-Distance Auxiliary Supervision for structural constraint learning. CAS-UNet was evaluated on five public histopathological nuclei datasets, including CoNIC, CPM17, MoNuSeg, NuInsSeg, and PanNuke. In the main benchmark comparison, CAS-UNet achieved the highest Dice and IoU values among the evaluated segmentation methods across the five datasets. Supplementary controlled experiments with five paired random seeds further demonstrated consistent improvements over U-Net and DeepLabV3+ under identical data partitions and training configurations. Additional experiments showed improved performance compared with traditional image-processing pipelines and more consistent segmentation results with Group Normalization than Batch Normalization under different batch sizes. Multi-seed component ablation experiments further characterized the contributions of PACP, NSSM, GN-based decoding, and BDAS. The proposed CAS-UNet provides an accurate and reliable framework for binary nuclei-region segmentation. Future work will focus on instance separation, stain-robust domain adaptation, independent clinical validation, and whole-slide image evaluation to further explore its potential for downstream pathological analysis.
- 8 August 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (5) , 156
LCV-YOLO: a lightweight and robust safety-helmet detection framework for complex underground mining environments
- 8 August 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (5) , 156
LCV-YOLO: a lightweight and robust safety-helmet detection framework for complex underground mining environments
- Top Cited Papers
- 10 February 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (2) , 61
YOLO-VCL: a lightweight and high-precision PCB defect detection network model
- 10 February 2026
- Published by Springer Nature in Journal of Real-Time Image Processing
- Vol. 23 (2) , 61
Top Cited PapersYOLO-VCL: a lightweight and high-precision PCB defect detection network model
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 5900-5912
TEDR-YNet: A Dual-Encoder Retinal Multi-Lesion Segmentation Network
Multi-lesion segmentation in diabetic retinopathy faces numerous challenges, including complex fundus structures, small inter-class differences between lesions, blurred lesion boundaries, and class imbalance. To address these challenges, this paper proposes TEDR-YNet, an innovative Y-shaped dual-encoder network for the segmentation of diabetic retinopathy lesions. This network utilizes a dual-encoder architecture to capture features from different representations, significantly enhancing the model’s expressive capability. In the bottleneck layer, a meticulously designed lesion-aware enhancement module is introduced to improve multi-scale lesion perception and boundary segmentation accuracy.Askip connection mechanism with gated convolution is innovatively incorporated to strengthen the model’s focus on lesion regions. An edge enhancement module is further employed to optimize boundary segmentation performance, and a hybrid loss function is designed to effectively mitigate the class imbalance problem. Extensive experiments conducted on three public datasets IDRID, DDR, and TJDR demonstrate the superior performance of TEDR-YNet, achieving average Dice coefficients of 68.61%, 58.51%, and 58.14%, respectively. These results represent improvements of 11.47%, 7.15%, and 9.8% over the baseline UNet model. Additionally, the mIOU metrics on the three datasets reached 53.24%, 46.84%, and 44.43%, respectively. These findings robustly demonstrate the precision and robustness of TEDR-YNet in multi-lesion segmentation for diabetic retinopathy, offering strong technical support for clinical diagnostic assistance.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 5900-5912
Open AccessTEDR-YNet: A Dual-Encoder Retinal Multi-Lesion Segmentation Network
Multi-lesion segmentation in diabetic retinopathy faces numerous challenges, including complex fundus structures, small inter-class differences between lesions, blurred lesion boundaries, and class imbalance. To address these challenges, this paper proposes TEDR-YNet, an innovative Y-shaped dual-encoder network for the segmentation of diabetic retinopathy lesions. This network utilizes a dual-encoder architecture to capture features from different representations, significantly enhancing the model’s expressive capability. In the bottleneck layer, a meticulously designed lesion-aware enhancement module is introduced to improve multi-scale lesion perception and boundary segmentation accuracy.Askip connection mechanism with gated convolution is innovatively incorporated to strengthen the model’s focus on lesion regions. An edge enhancement module is further employed to optimize boundary segmentation performance, and a hybrid loss function is designed to effectively mitigate the class imbalance problem. Extensive experiments conducted on three public datasets IDRID, DDR, and TJDR demonstrate the superior performance of TEDR-YNet, achieving average Dice coefficients of 68.61%, 58.51%, and 58.14%, respectively. These results represent improvements of 11.47%, 7.15%, and 9.8% over the baseline UNet model. Additionally, the mIOU metrics on the three datasets reached 53.24%, 46.84%, and 44.43%, respectively. These findings robustly demonstrate the precision and robustness of TEDR-YNet in multi-lesion segmentation for diabetic retinopathy, offering strong technical support for clinical diagnostic assistance.
- 29 November 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (17) , 1610
PCTUNet: A Parallel CNN-Transformer U-Net for High-Accuracy Spectral Reconstruction from RGB Images
- 29 November 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (17) , 1610
PCTUNet: A Parallel CNN-Transformer U-Net for High-Accuracy Spectral Reconstruction from RGB Images
- Open Access
- 18 August 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (7) , 1-16
Retinal vascular segmentation network based on dual-scale morphological enhancement
- 18 August 2025
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 37 (7) , 1-16
Open AccessRetinal vascular segmentation network based on dual-scale morphological enhancement
- Open Access
- 19 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 144174-144185
TrUNet: Dual-Branch Network by Fusing CNN and Transformer for Skin Lesion Segmentation
In the medical field, precise segmentation of skin lesion areas is essential for accurate diagnosis and treatment of diseases. Due to the varied morphologies and fuzzy boundaries of skin lesions, as well as interference from hair coverage, segmentation tasks are extremely challenging. To address the problem, a network called TrUNet is proposed, which combines the advantages of Transformer and convolutional neural networks (CNNs). Transformer and Res2Net are taken as two branches of the encoder in this network, with the goal of extracting rich global information for precise lesion segmentation in medical images. Firstly, the TrFusion module was designed to selectively fuse complementary features extracted by the Transformer branch and the Res2Net branch in the encoder, enhancing important information while suppressing irrelevant details. Secondly, the Multi-Scale Feature Aggregation (MFA) module was designed to fuse feature representations from different stages of the same branch to complement positional and spatial information. Finally, to validate the effectiveness of the proposed method, experiments were conducted on the ISIC2017, ISIC2018, and PH2 datasets. TrUNet achieved Dice coefficient of 90.61%, IoU of 84.25%, and Accuracy of 94.74% on the ISIC2018 dataset. This indicates that our model has enormous potential in the field of medical image segmentation.
- 19 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 144174-144185
Open AccessTrUNet: Dual-Branch Network by Fusing CNN and Transformer for Skin Lesion Segmentation
In the medical field, precise segmentation of skin lesion areas is essential for accurate diagnosis and treatment of diseases. Due to the varied morphologies and fuzzy boundaries of skin lesions, as well as interference from hair coverage, segmentation tasks are extremely challenging. To address the problem, a network called TrUNet is proposed, which combines the advantages of Transformer and convolutional neural networks (CNNs). Transformer and Res2Net are taken as two branches of the encoder in this network, with the goal of extracting rich global information for precise lesion segmentation in medical images. Firstly, the TrFusion module was designed to selectively fuse complementary features extracted by the Transformer branch and the Res2Net branch in the encoder, enhancing important information while suppressing irrelevant details. Secondly, the Multi-Scale Feature Aggregation (MFA) module was designed to fuse feature representations from different stages of the same branch to complement positional and spatial information. Finally, to validate the effectiveness of the proposed method, experiments were conducted on the ISIC2017, ISIC2018, and PH2 datasets. TrUNet achieved Dice coefficient of 90.61%, IoU of 84.25%, and Accuracy of 94.74% on the ISIC2018 dataset. This indicates that our model has enormous potential in the field of medical image segmentation.
- Open Access
- 11 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 52555-52564
VB-SOLO: Single-Stage Instance Segmentation of Overlapping Epithelial Cells
The instance segmentation of overlapping cells in smear images of epithelial cells is challenging due to the significant overlap and adhesion between the cells’ translucent cytoplasm. In this paper, an improved single-stage instance segmentation network called VoVNet-BiFPN-SOLO (VB-SOLO) is proposed to address this problem. The model takes SOLOv2 model as its main frame. Firstly, the backbone network uses Efficient Channel Attention (ECA) to optimize the VoVNetv2 network to increase the information interaction across channels and enhance the extraction of cell instance features. Secondly, the bi-directional feature pyramid network (BiFPN) is introduced to connect with the new backbone. BiFPN can achieve the weighted fusion of features with different resolutions from bottom to top and keep more shallow semantic information in the network. Finally, the Convolutional Block Attention Module (CBAM) is added to the mask branch to improve cell segmentation results in feature maps. Experimental results on the publicly available datasets CISD and Cx22 demonstrate the effectiveness of the VB-SOLO model, achieving a DCP of 0.966 and 0.940 and a FNRO of 0.055 and 0.03. Compared to the original SOLOv2 algorithm, the proposed method achieved improvements in DCP of 1.3% and 1.1% respectively. Additionally, comparative tests with multiple instance segmentation networks have shown that the proposed improved network can achieve a better balance between segmentation accuracy and efficiency. The experimental results demonstrate the effectiveness of the proposed network improvements and the potential of single-stage instance segmentation networks in overlapping cell image segmentation.
- 11 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 52555-52564
Open AccessVB-SOLO: Single-Stage Instance Segmentation of Overlapping Epithelial Cells
The instance segmentation of overlapping cells in smear images of epithelial cells is challenging due to the significant overlap and adhesion between the cells’ translucent cytoplasm. In this paper, an improved single-stage instance segmentation network called VoVNet-BiFPN-SOLO (VB-SOLO) is proposed to address this problem. The model takes SOLOv2 model as its main frame. Firstly, the backbone network uses Efficient Channel Attention (ECA) to optimize the VoVNetv2 network to increase the information interaction across channels and enhance the extraction of cell instance features. Secondly, the bi-directional feature pyramid network (BiFPN) is introduced to connect with the new backbone. BiFPN can achieve the weighted fusion of features with different resolutions from bottom to top and keep more shallow semantic information in the network. Finally, the Convolutional Block Attention Module (CBAM) is added to the mask branch to improve cell segmentation results in feature maps. Experimental results on the publicly available datasets CISD and Cx22 demonstrate the effectiveness of the VB-SOLO model, achieving a DCP of 0.966 and 0.940 and a FNRO of 0.055 and 0.03. Compared to the original SOLOv2 algorithm, the proposed method achieved improvements in DCP of 1.3% and 1.1% respectively. Additionally, comparative tests with multiple instance segmentation networks have shown that the proposed improved network can achieve a better balance between segmentation accuracy and efficiency. The experimental results demonstrate the effectiveness of the proposed network improvements and the potential of single-stage instance segmentation networks in overlapping cell image segmentation.
- Open Access
- 1 April 2023
- Published by SPIE-Intl Soc Optical Eng in Journal of Biomedical Optics
- Vol. 28 (04) , 045002-16
Improved U-net-based leukocyte segmentation method
Leukocytes are an important part of the body’s immune system. They are mainly made up of neutrophils, basophils, eosinophils, monocytes, and lymphocytes. The number and ratio of different types of leukocytes correspond to different diseases, so the accuracy of the leukocyte test is very important for the diagnosis of diseases. The traditional method of leukocyte testing is mainly manual microscopy, which requires two well-trained and experienced examiners to observe the morphology of leukocytes under the microscope to finalize the test results, which is not only time-consuming and inefficient but also highly subjective.1 With the improvement of computer performance and the increase of data volume in recent years, deep learning has also been widely applied to medical segmentation, and the proposed work is mainly based on various convolutional neural network schemes,2 such as SegNet,3,4 U-Net,4,5 and VGG-Unet.6 For the work on leukocyte segmentation, many researchers have achieved good results. Reena et al.7 used DeepLabv3+ architecture and ResNet-50 as a feature extraction network to segment each of the five types of leukocytes, and the final average segmentation accuracy was 96.1% and intersection-union accuracy was 92.1%; Lu et al.8 proposed a WBC-Net model based on UNet++ and ResNet to segment individual leukocytes, mainly by designing a context-aware feature encoder with residual blocks to extract multiscale features and introducing hybrid jump paths on dense convolutional blocks to acquire and fuse image features at different scales, which finally achieved good results on four publicly available datasets. Li et al.9 proposed a segmentation algorithm based on the U-Net model with dual-path and void space pyramidal pooling to achieve pixel-level segmentation of blood leukocytes, mainly by introducing a dual-path network in the feature encoder to extract multiscale features of leukocytes in images and using a void space pyramidal pooling module to enhance the feature extraction capability of the network, and the research results show that its mIoU value can reach above 0.97.
- 1 April 2023
- Published by SPIE-Intl Soc Optical Eng in Journal of Biomedical Optics
- Vol. 28 (04) , 045002-16
Open AccessImproved U-net-based leukocyte segmentation method
Leukocytes are an important part of the body’s immune system. They are mainly made up of neutrophils, basophils, eosinophils, monocytes, and lymphocytes. The number and ratio of different types of leukocytes correspond to different diseases, so the accuracy of the leukocyte test is very important for the diagnosis of diseases. The traditional method of leukocyte testing is mainly manual microscopy, which requires two well-trained and experienced examiners to observe the morphology of leukocytes under the microscope to finalize the test results, which is not only time-consuming and inefficient but also highly subjective.1 With the improvement of computer performance and the increase of data volume in recent years, deep learning has also been widely applied to medical segmentation, and the proposed work is mainly based on various convolutional neural network schemes,2 such as SegNet,3,4 U-Net,4,5 and VGG-Unet.6 For the work on leukocyte segmentation, many researchers have achieved good results. Reena et al.7 used DeepLabv3+ architecture and ResNet-50 as a feature extraction network to segment each of the five types of leukocytes, and the final average segmentation accuracy was 96.1% and intersection-union accuracy was 92.1%; Lu et al.8 proposed a WBC-Net model based on UNet++ and ResNet to segment individual leukocytes, mainly by designing a context-aware feature encoder with residual blocks to extract multiscale features and introducing hybrid jump paths on dense convolutional blocks to acquire and fuse image features at different scales, which finally achieved good results on four publicly available datasets. Li et al.9 proposed a segmentation algorithm based on the U-Net model with dual-path and void space pyramidal pooling to achieve pixel-level segmentation of blood leukocytes, mainly by introducing a dual-path network in the feature encoder to extract multiscale features of leukocytes in images and using a void space pyramidal pooling module to enhance the feature extraction capability of the network, and the research results show that its mIoU value can reach above 0.97.