Yushi Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yushi Liu has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Yan Yan
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China
17 shared publications - Mingyang Hou
School of Microelectronics and Communication Engineering, Chongqing University, Chongqing, 400044, China
12 shared publications - Zhi Yu
School of Microelectronics and Communication Engineering, Chongqing University, Chongqing, China
12 shared publications - Zhiyong Huang
School of Microelectronics and Communication Engineering, Chongqing University, Chongqing, 400044, China; Key Laboratory of Dependable Service Computing in Cyber Physical Society (Chongqing University), Ministry of Education of China, Chongqing University, Chongqing, 400044, China
11 shared publications - Jiahong Wang
The Second Medical Center, Health Management Institute, National Clinical Research Center for Geriatric Diseases, Chinese PLA General Hospital, Beijing, 100853, China
7 shared publications - Xiaoyu Li
Bioengineering College, Chongqing University, Chongqing, China
6 shared publications
Latest Publications
- 12 August 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 351
Bridging teacher–student representation domains for manifold-aware knowledge distillation
- 12 August 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 351
Bridging teacher–student representation domains for manifold-aware knowledge distillation
- 15 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Biomedical Engineering
- Vol. PP (00189294) , 1-12
Mapper-Based Topological EEG Modeling for Task Representation in Robot-Assisted Surgery
Robot-assisted surgery (RAS) requires objective and interpretable frameworks for task representation and analysis. Although electroencephalography (EEG) has been increasingly used to capture neural correlates of surgical behavior, most existing methods rely on point-wise classification or low-dimensional geometric embeddings, offering limited insight into the organization of task-related neural representations. To address this limitation, a Mapper-based topological EEG modeling framework is introduced to characterize task-related neural state spaces during robot-assisted surgical tasks. Using EEG recordings from an RAS dataset covering both coarse- and fine-grained surgical tasks, phase-locking value (PLV) connectivity features are extracted across canonical frequency bands, and Mapper is used to model the organization and transitions of task-related neural states. Persistent homology and BrainNetCNN are also incorporated as additional structural and global modeling approaches for a more balanced comparison. Within-subject analyses show that the proposed framework consistently captures coherent and separable task-related structures, with frequency-dependent neural topology enhancing task discrimination, particularly in higher-frequency bands. In six-class task modeling, an average accuracy of 94.78% is achieved. Comparative evaluation against embedding- and clustering-based baselines shows that Mapper provides stronger structural consistency and clearer task-related organization. In cross-subject settings, however, substantial inter-subject variability remains, indicating that task-related neural topology is strongly individualized. These results indicate that Mapper-based topological modeling provides an interpretable foundation for task representation and personalized EEG-based analysis in RAS. This framework may further inform future individualized monitoring of task execution and training analysis in robot-assisted surgery.
- 15 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Biomedical Engineering
- Vol. PP (00189294) , 1-12
Mapper-Based Topological EEG Modeling for Task Representation in Robot-Assisted Surgery
Robot-assisted surgery (RAS) requires objective and interpretable frameworks for task representation and analysis. Although electroencephalography (EEG) has been increasingly used to capture neural correlates of surgical behavior, most existing methods rely on point-wise classification or low-dimensional geometric embeddings, offering limited insight into the organization of task-related neural representations. To address this limitation, a Mapper-based topological EEG modeling framework is introduced to characterize task-related neural state spaces during robot-assisted surgical tasks. Using EEG recordings from an RAS dataset covering both coarse- and fine-grained surgical tasks, phase-locking value (PLV) connectivity features are extracted across canonical frequency bands, and Mapper is used to model the organization and transitions of task-related neural states. Persistent homology and BrainNetCNN are also incorporated as additional structural and global modeling approaches for a more balanced comparison. Within-subject analyses show that the proposed framework consistently captures coherent and separable task-related structures, with frequency-dependent neural topology enhancing task discrimination, particularly in higher-frequency bands. In six-class task modeling, an average accuracy of 94.78% is achieved. Comparative evaluation against embedding- and clustering-based baselines shows that Mapper provides stronger structural consistency and clearer task-related organization. In cross-subject settings, however, substantial inter-subject variability remains, indicating that task-related neural topology is strongly individualized. These results indicate that Mapper-based topological modeling provides an interpretable foundation for task representation and personalized EEG-based analysis in RAS. This framework may further inform future individualized monitoring of task execution and training analysis in robot-assisted surgery.
- 11 March 2026
- Published by Elsevier in Medical Image Analysis
- Vol. 111, 104037
Eliminating domain-related confounding factors in cross-domain one-shot medical image segmentation via causal inference
- 11 March 2026
- Published by Elsevier in Medical Image Analysis
- Vol. 111, 104037
Eliminating domain-related confounding factors in cross-domain one-shot medical image segmentation via causal inference
- 21 February 2026
- Published by Elsevier in Pattern Recognition
- Vol. 177
Enhancing Radiography-Report foundation model via Multi-View masked contrastive learning
- 21 February 2026
- Published by Elsevier in Pattern Recognition
- Vol. 177
Enhancing Radiography-Report foundation model via Multi-View masked contrastive learning
- 26 December 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 335
BACFormer: A robust boundary-aware transformer for medical image segmentation
- 26 December 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 335
BACFormer: A robust boundary-aware transformer for medical image segmentation
- 13 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 5301-5314
Overcoming Limitations in One-Shot Semantic Segmentation via Adaptive Visual-Text Guided Prototype Relationship Optimization
The scarcity of high-quality annotated data in medical imaging significantly constrains the performance of deep learning-based segmentation models. While few-shot medical image segmentation (FSMIS) has emerged as a promising solution, existing methods exhibit critical limitations when handling scenarios with inter-class similarity between foreground-background regions and intra-class heterogeneity within foreground objects. Current prototype-based approaches focus primarily on the holistic extraction of the prototype from support images, failing to distinguish subtle anatomical variations and complex feature representations effectively. The AVT-ProNet features three innovative components: 1) An Adaptive Visual-Text Prototype Generation (AVPG) module leveraging CLIP’s cross-modal guide capabilities through adaptive prompting strategies; 2) a graph-based multiregion prototyping relationship optimization (GMPRO) module establishing structural relationships between decomposed subregion prototypes via graph neural networks; 3) a foreground-background prototyping contrast learning (FBPCL) strategy implementing dual-space optimization through inter-class separation and intra-class compactness. The synergistic integration of multi-modal guidance, structural relationship modeling, and contrastive prototype refinement enables our framework to overcome existing limitations in FSMIS. Comprehensive evaluations across multiple clinical scenarios (CHAOS, SABS, and CMR datasets under diverse training configurations) demonstrate superior performance over state-of-the-art approaches, including PANet, CAT-Net, DMAP, and recent PAMI baselines, Source code is available at https://github.com/394481125/AVT-ProNet.
- 13 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 5301-5314
Overcoming Limitations in One-Shot Semantic Segmentation via Adaptive Visual-Text Guided Prototype Relationship Optimization
The scarcity of high-quality annotated data in medical imaging significantly constrains the performance of deep learning-based segmentation models. While few-shot medical image segmentation (FSMIS) has emerged as a promising solution, existing methods exhibit critical limitations when handling scenarios with inter-class similarity between foreground-background regions and intra-class heterogeneity within foreground objects. Current prototype-based approaches focus primarily on the holistic extraction of the prototype from support images, failing to distinguish subtle anatomical variations and complex feature representations effectively. The AVT-ProNet features three innovative components: 1) An Adaptive Visual-Text Prototype Generation (AVPG) module leveraging CLIP’s cross-modal guide capabilities through adaptive prompting strategies; 2) a graph-based multiregion prototyping relationship optimization (GMPRO) module establishing structural relationships between decomposed subregion prototypes via graph neural networks; 3) a foreground-background prototyping contrast learning (FBPCL) strategy implementing dual-space optimization through inter-class separation and intra-class compactness. The synergistic integration of multi-modal guidance, structural relationship modeling, and contrastive prototype refinement enables our framework to overcome existing limitations in FSMIS. Comprehensive evaluations across multiple clinical scenarios (CHAOS, SABS, and CMR datasets under diverse training configurations) demonstrate superior performance over state-of-the-art approaches, including PANet, CAT-Net, DMAP, and recent PAMI baselines, Source code is available at https://github.com/394481125/AVT-ProNet.
- 22 July 2025
- Published by Elsevier in Computerized Medical Imaging and Graphics
- Vol. 124, 102610
Multi-scale interaction and locally enhanced bridging network for medical image segmentation
- 22 July 2025
- Published by Elsevier in Computerized Medical Imaging and Graphics
- Vol. 124, 102610
Multi-scale interaction and locally enhanced bridging network for medical image segmentation
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (10) , 7312-7325
MSCD-VM-UNet: A Vision Mamba Combining Multi-Scale Global and Local Feature Extraction With Cross-Domain Feature Fusion for Medical Image Segmentation
Accurate segmentation of tissues and lesions is essential for diagnosis and treatment. State Space Models (SSMs) have gained attention for their linear complexity and ability to model long-range dependencies. However, the existing Mamba architecture relies on direct skip connections, which limits its ability to integrate multi-scale and multi-level features and handle boundary details effectively. To address these limitations, we propose the MSCD-VM-UNet architecture, which incorporates three novel modules: the Spatial Group Multi-Scale Attention Module (SGMAM), the Cross-Domain Feature Fusion Module (CDFFM), and the Attention-Based Feature Injection Module (ABFIM). The SGMAM captures multi-scale global and local information and adaptively adjusts feature importance to highlight key regions while suppressing noise. The CDFFM enhances boundary and detail handling by aligning semantic features from both the frequency and spatial domains. The ABFIM utilizes attention mechanisms to adaptively fuse and weigh features from different scales and semantics, promoting feature collaboration and improving the model's robustness in complex tasks. Experiments on multiple datasets show that these modules significantly enhance the accuracy of MSCD-VM-UNet, setting a new benchmark for medical image segmentation. Our code will be made available at https://github.com/StphenWang/MSCD-VM-UNet.
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (10) , 7312-7325
MSCD-VM-UNet: A Vision Mamba Combining Multi-Scale Global and Local Feature Extraction With Cross-Domain Feature Fusion for Medical Image Segmentation
Accurate segmentation of tissues and lesions is essential for diagnosis and treatment. State Space Models (SSMs) have gained attention for their linear complexity and ability to model long-range dependencies. However, the existing Mamba architecture relies on direct skip connections, which limits its ability to integrate multi-scale and multi-level features and handle boundary details effectively. To address these limitations, we propose the MSCD-VM-UNet architecture, which incorporates three novel modules: the Spatial Group Multi-Scale Attention Module (SGMAM), the Cross-Domain Feature Fusion Module (CDFFM), and the Attention-Based Feature Injection Module (ABFIM). The SGMAM captures multi-scale global and local information and adaptively adjusts feature importance to highlight key regions while suppressing noise. The CDFFM enhances boundary and detail handling by aligning semantic features from both the frequency and spatial domains. The ABFIM utilizes attention mechanisms to adaptively fuse and weigh features from different scales and semantics, promoting feature collaboration and improving the model's robustness in complex tasks. Experiments on multiple datasets show that these modules significantly enhance the accuracy of MSCD-VM-UNet, setting a new benchmark for medical image segmentation. Our code will be made available at https://github.com/StphenWang/MSCD-VM-UNet.
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
MCECF: A Multiscale Complementary Enhanced Context Fusion Network for Remote Sensing Change Detection
Remote sensing change detection (RSCD) holds significant research value in remote sensing image processing. In recent years, many researchers have achieved remarkable results in RSCD tasks using methods based on CNNs or Transformers. Considering the limited receptive field of CNN models and the high computational cost of Transformers, many researchers have combined the two approaches, yielding promising results. However, most current RSCD-based models focus solely on change and temporal information, overlooking their complementary relationship. Additionally, some multi-scale feature fusion methods emphasize enhancing individual scales while neglecting the correlations between different scales. To address the above issues, we propose a multi-scale complementary enhanced context fusion network (MCECF). The network first introduces a global-local context aggregation module (GLCAM) to capture global-local context information while extracting multi-level feature maps. Subsequently, a complementary enhancement difference module (CEDM) is employed to complementarily aggregate the captured change and temporal information of bi-temporal remote sensing image features. To fully leverage the correlations between multi-scale features, a progressive decoder comprising a supervised spatial attention (SSA) mechanism and a multi-scale complementary enhanced fusion module (MCEFM) was developed. Moreover, to tackle the disparity between changed and unchanged regions, a dual-branch dynamic attention fusion module (DAFM) was designed to enhance the model’s adaptability to diverse scenarios. We conducted comparative experiments on five RSCD datasets against nine SOTA methods, and the results confirmed the effectiveness of the proposed MCECF in RSCD tasks. Our code will be made available at https://github.com/kakuqikaduo/MCECF.
- 1 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-14
MCECF: A Multiscale Complementary Enhanced Context Fusion Network for Remote Sensing Change Detection
Remote sensing change detection (RSCD) holds significant research value in remote sensing image processing. In recent years, many researchers have achieved remarkable results in RSCD tasks using methods based on CNNs or Transformers. Considering the limited receptive field of CNN models and the high computational cost of Transformers, many researchers have combined the two approaches, yielding promising results. However, most current RSCD-based models focus solely on change and temporal information, overlooking their complementary relationship. Additionally, some multi-scale feature fusion methods emphasize enhancing individual scales while neglecting the correlations between different scales. To address the above issues, we propose a multi-scale complementary enhanced context fusion network (MCECF). The network first introduces a global-local context aggregation module (GLCAM) to capture global-local context information while extracting multi-level feature maps. Subsequently, a complementary enhancement difference module (CEDM) is employed to complementarily aggregate the captured change and temporal information of bi-temporal remote sensing image features. To fully leverage the correlations between multi-scale features, a progressive decoder comprising a supervised spatial attention (SSA) mechanism and a multi-scale complementary enhanced fusion module (MCEFM) was developed. Moreover, to tackle the disparity between changed and unchanged regions, a dual-branch dynamic attention fusion module (DAFM) was designed to enhance the model’s adaptability to diverse scenarios. We conducted comparative experiments on five RSCD datasets against nine SOTA methods, and the results confirmed the effectiveness of the proposed MCECF in RSCD tasks. Our code will be made available at https://github.com/kakuqikaduo/MCECF.