Shenwen Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shenwen Wang has a total of 46 co-authors. Here are the top co-authors by number of shared publications:
- Zhaolu Guo
Institute of Medical Informatics and Engineering, School of Science, JiangXi University of Science and Technology, Ganzhou, China
5 shared publications Zhenfeng Zhao
Hebei Huiji Technology Development Co., Ltd. Shijiazhuang Hebei China
4 shared publicationsQi Liu
School of Information Engineering, Hebei GEO University, Shijiazhuang, China
3 shared publications- Zhenfeng Zhao
Hebei Huiji Technology Development Co., Ltd, Shijiazhuang, 050000, China
3 shared publications - Feng Yang
School of Information Engineering, Hebei GEO University, Shijiazhuang, China
2 shared publications Haibin Wang
2 shared publications
Latest Publications
- 10 January 2026
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 36 (1)
MorphoFormer : Dual‐Branch Dilated Transformer With Pathological Prior Fusion for Cervical Cell Morphology Analysis
Cervical cancer is one of the most common malignant tumors among women worldwide, and accurate early diagnosis is critical for improving patient survival rates. Traditional cytological screening methods rely on manual microscopic examination, which suffers from low efficiency and high subjectivity. In recent years, deep learning has facilitated the automation of cervical cell image analysis, yet challenges such as insufficient modeling of pathological features and high computational cost remain. To address these issues, this study proposes a novel dual‐branch multi‐scale model, MorphoFormer. The model employs a multi‐scale dilated Transformer (DilateFormer) as its backbone and innovatively incorporates specialized modules for each branch: A Local Context Aggregation (LCA) module in the local branch and a Global Focus Attention (GFA) module in the global branch. These modules respectively enhance the representation of local details and global semantics, and their features are fused to enable collaborative multi‐scale information modeling. Experimental results on the publicly available SIPaKMeD dataset demonstrate that MorphoFormer achieves classification accuracies of 99.58%, 98.51%, and 98.14% for binary, three‐class, and five‐class tasks, respectively. Further validation on the Blood Cell Count and Detection (BCCD) dataset indicates strong cross‐task robustness. Moreover, MorphoFormer requires only 8.22 GFLOPs for inference, highlighting its practical potential by achieving high performance with low computational overhead. Related codes: https://github.com/sijhb/MorphoFormer .
- 10 January 2026
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 36 (1)
MorphoFormer : Dual‐Branch Dilated Transformer With Pathological Prior Fusion for Cervical Cell Morphology Analysis
Cervical cancer is one of the most common malignant tumors among women worldwide, and accurate early diagnosis is critical for improving patient survival rates. Traditional cytological screening methods rely on manual microscopic examination, which suffers from low efficiency and high subjectivity. In recent years, deep learning has facilitated the automation of cervical cell image analysis, yet challenges such as insufficient modeling of pathological features and high computational cost remain. To address these issues, this study proposes a novel dual‐branch multi‐scale model, MorphoFormer. The model employs a multi‐scale dilated Transformer (DilateFormer) as its backbone and innovatively incorporates specialized modules for each branch: A Local Context Aggregation (LCA) module in the local branch and a Global Focus Attention (GFA) module in the global branch. These modules respectively enhance the representation of local details and global semantics, and their features are fused to enable collaborative multi‐scale information modeling. Experimental results on the publicly available SIPaKMeD dataset demonstrate that MorphoFormer achieves classification accuracies of 99.58%, 98.51%, and 98.14% for binary, three‐class, and five‐class tasks, respectively. Further validation on the Blood Cell Count and Detection (BCCD) dataset indicates strong cross‐task robustness. Moreover, MorphoFormer requires only 8.22 GFLOPs for inference, highlighting its practical potential by achieving high performance with low computational overhead. Related codes: https://github.com/sijhb/MorphoFormer .
- Open Access
- 21 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1)
Multi scale deep learning quantifies Ki67 index in breast cancer histopathology images
The Ki67 proliferation index (PI) serves as a crucial prognostic indicator in clinical settings, widely utilized for evaluating breast cancer progression and forecasting chemotherapy efficacy. Nonetheless, conventional manual PI estimation methods are plagued by high subjectivity, time inefficiency, and limited reproducibility. Despite notable advancements in deep learning for Ki67 PI computation, current approaches face challenges in concurrently capturing local cellular morphological features and global tumor proliferation data, particularly when addressing issues like complex backgrounds, staining variability, tumor heterogeneity, and noise in low-resolution images. In response, we introduce Kpi-Net, a multi scale deep learning framework built on U-Net, designed to precisely quantify the Ki67 index in breast cancer pathology images. Initially, we develop a Residual Dilated Multi Scale Module (RDMS Module) that utilizes multi-branch dilated convolutions and residual connections to capture both local and global information, while integrating a Transformer Block to bolster global modeling, effectively mitigating missed and erroneous detections arising from irregular cell distribution. Next, we introduce the High-level Screening-Convolutional Block Attention Module Feature Pyramid Networks (HS-CBAM-FPN), incorporating channel and spatial attention mechanisms alongside the Selective Feature Fusion with CBAM (SFF-CBAM) to facilitate effective integration of multi-level features. Lastly, we apply the watershed algorithm to feature maps, refining cell cluster segmentation via distance transformation and local maximum detection, which improves the precision of Ki67 index computation. Comprehensive experimental results indicate that Kpi-Net surpasses current leading methods in metrics like F1 score and root mean square error (RMSE), highlighting its potential for accurate Ki67 index computation and precise cell detection, thereby offering a dependable tool for accurate breast cancer diagnosis and therapeutic decision-making.
- 21 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1)
Open AccessMulti scale deep learning quantifies Ki67 index in breast cancer histopathology images
The Ki67 proliferation index (PI) serves as a crucial prognostic indicator in clinical settings, widely utilized for evaluating breast cancer progression and forecasting chemotherapy efficacy. Nonetheless, conventional manual PI estimation methods are plagued by high subjectivity, time inefficiency, and limited reproducibility. Despite notable advancements in deep learning for Ki67 PI computation, current approaches face challenges in concurrently capturing local cellular morphological features and global tumor proliferation data, particularly when addressing issues like complex backgrounds, staining variability, tumor heterogeneity, and noise in low-resolution images. In response, we introduce Kpi-Net, a multi scale deep learning framework built on U-Net, designed to precisely quantify the Ki67 index in breast cancer pathology images. Initially, we develop a Residual Dilated Multi Scale Module (RDMS Module) that utilizes multi-branch dilated convolutions and residual connections to capture both local and global information, while integrating a Transformer Block to bolster global modeling, effectively mitigating missed and erroneous detections arising from irregular cell distribution. Next, we introduce the High-level Screening-Convolutional Block Attention Module Feature Pyramid Networks (HS-CBAM-FPN), incorporating channel and spatial attention mechanisms alongside the Selective Feature Fusion with CBAM (SFF-CBAM) to facilitate effective integration of multi-level features. Lastly, we apply the watershed algorithm to feature maps, refining cell cluster segmentation via distance transformation and local maximum detection, which improves the precision of Ki67 index computation. Comprehensive experimental results indicate that Kpi-Net surpasses current leading methods in metrics like F1 score and root mean square error (RMSE), highlighting its potential for accurate Ki67 index computation and precise cell detection, thereby offering a dependable tool for accurate breast cancer diagnosis and therapeutic decision-making.
- 6 November 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 113
DBFRNet: A dual-branch feature refinement network based on Pyramid Vision Transformer for polyp segmentation
- 6 November 2025
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 113
DBFRNet: A dual-branch feature refinement network based on Pyramid Vision Transformer for polyp segmentation
- 1 October 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (15) , 1-56
Two-Arch2-DAC: a two-stage DAV-$I_{\varepsilon +}$ cooperative framework for MaOPs- 1 October 2025
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 81 (15) , 1-56
Two-Arch2-DAC: a two-stage DAV-$I_{\varepsilon +}$ cooperative framework for MaOPs- Open Access
- 7 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-19
Multi-module UNet++ for colon cancer histopathological image segmentation
In the pathological diagnosis of colorectal cancer, the precise segmentation of glandular and cellular contours serves as the fundamental basis for achieving accurate clinical diagnosis. However, this task presents significant challenges due to complex phenomena such as nuclear staining heterogeneity, variations in nuclear size, boundary overlap, and nuclear clustering. With the continuous advancement of deep learning techniques—particularly encoder-decoder architectures—and the emergence of various high-performance functional modules, multi module collaborative fusion has become an effective approach to enhance segmentation performance. To this end, this study proposes the RPAU-Net++ model, which integrates the ResNet-50 encoder (R), the Joint Pyramid Fusion Module (P), and the Convolutional Block Attention Module (A) into the UNet++ framework, forming a multi-module-enhanced segmentation architecture. Specifically, ResNet-50 mitigates gradient vanishing and degradation issues in deep network training through residual skip connections, thereby improving model convergence stability and feature representation depth. JPFM achieves progressive fusion of cross-layer features via a multi-scale feature pyramid, enhancing the encoding capability for complex tissue structures and fine boundary information. CBAM employs adaptive weight allocation in both spatial and channel dimensions to focus on target region features while effectively suppressing irrelevant background noise, thereby improving feature discriminability. Comparative experiments on the GlaS and CoNIC colorectal cancer pathology datasets, as well as the more challenging PanNuke dataset, demonstrate that RPAU-Net++ significantly outperforms mainstream models in key segmentation metrics such as IoU and Dice, providing a more accurate solution for pathological image segmentation in colorectal cancer.
- 7 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-19
Open AccessMulti-module UNet++ for colon cancer histopathological image segmentation
In the pathological diagnosis of colorectal cancer, the precise segmentation of glandular and cellular contours serves as the fundamental basis for achieving accurate clinical diagnosis. However, this task presents significant challenges due to complex phenomena such as nuclear staining heterogeneity, variations in nuclear size, boundary overlap, and nuclear clustering. With the continuous advancement of deep learning techniques—particularly encoder-decoder architectures—and the emergence of various high-performance functional modules, multi module collaborative fusion has become an effective approach to enhance segmentation performance. To this end, this study proposes the RPAU-Net++ model, which integrates the ResNet-50 encoder (R), the Joint Pyramid Fusion Module (P), and the Convolutional Block Attention Module (A) into the UNet++ framework, forming a multi-module-enhanced segmentation architecture. Specifically, ResNet-50 mitigates gradient vanishing and degradation issues in deep network training through residual skip connections, thereby improving model convergence stability and feature representation depth. JPFM achieves progressive fusion of cross-layer features via a multi-scale feature pyramid, enhancing the encoding capability for complex tissue structures and fine boundary information. CBAM employs adaptive weight allocation in both spatial and channel dimensions to focus on target region features while effectively suppressing irrelevant background noise, thereby improving feature discriminability. Comparative experiments on the GlaS and CoNIC colorectal cancer pathology datasets, as well as the more challenging PanNuke dataset, demonstrate that RPAU-Net++ significantly outperforms mainstream models in key segmentation metrics such as IoU and Dice, providing a more accurate solution for pathological image segmentation in colorectal cancer.
- 25 March 2025
- Published by Springer Nature in Applied Intelligence
- Vol. 55 (7) , 1-16
ResU-KAN: a medical image segmentation model integrating residual convolutional attention and atrous spatial pyramid pooling
- 25 March 2025
- Published by Springer Nature in Applied Intelligence
- Vol. 55 (7) , 1-16
ResU-KAN: a medical image segmentation model integrating residual convolutional attention and atrous spatial pyramid pooling
- 23 November 2024
- Published by Springer Nature in Applied Intelligence
- Vol. 55 (1) , 1-17
PVT-MA: pyramid vision transformers with multi-attention fusion mechanism for polyp segmentation
- 23 November 2024
- Published by Springer Nature in Applied Intelligence
- Vol. 55 (1) , 1-17
PVT-MA: pyramid vision transformers with multi-attention fusion mechanism for polyp segmentation
- 7 February 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (11) , 12470-12496
Multimodal and multi-objective optimization algorithm based on two-stage search framework
- 7 February 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (11) , 12470-12496
Multimodal and multi-objective optimization algorithm based on two-stage search framework
- 10 May 2021
- Published by Inderscience Publishers in International Journal of Bio-Inspired Computation
- Vol. 17 (3) , 154
Improved gravitational search algorithm based on chaotic local search
The traditional gravitational search algorithm (GSA) maintains good diversity of solutions but often demonstrates weak local search ability. To promote the local search ability of GSA, a new GSA based on chaotic local search (CLSGSA) is introduced in this paper. In its search operations, CLSGSA first executes the conventional search operations of the basic GSA to maintain the diversity of solutions. After that, CLSGSA executes a chaotic local search with the search experience from the current best solution to increase the local search capability. In the experiments, we utilise a suite of benchmark functions to verify the performance of CLSGSA. Moreover, we compare the proposed CLSGSA with several GSA variants. The comparisons validate the effectiveness of CLSGSA.
- 10 May 2021
- Published by Inderscience Publishers in International Journal of Bio-Inspired Computation
- Vol. 17 (3) , 154
Improved gravitational search algorithm based on chaotic local search
The traditional gravitational search algorithm (GSA) maintains good diversity of solutions but often demonstrates weak local search ability. To promote the local search ability of GSA, a new GSA based on chaotic local search (CLSGSA) is introduced in this paper. In its search operations, CLSGSA first executes the conventional search operations of the basic GSA to maintain the diversity of solutions. After that, CLSGSA executes a chaotic local search with the search experience from the current best solution to increase the local search capability. In the experiments, we utilise a suite of benchmark functions to verify the performance of CLSGSA. Moreover, we compare the proposed CLSGSA with several GSA variants. The comparisons validate the effectiveness of CLSGSA.
- 9 January 2021
- Published by Springer Nature in Applied Intelligence
- Vol. 51 (8) , 5525-5542
A new dominance relation based on convergence indicators and niching for many-objective optimization
- 9 January 2021
- Published by Springer Nature in Applied Intelligence
- Vol. 51 (8) , 5525-5542
A new dominance relation based on convergence indicators and niching for many-objective optimization