Clinically-Grounded Counterfactual Reasoning for Medical Video Diagnosis

Jianzhe Gao, Churan Wang✉, Weiyi Zhang, Jianghua Li, Li-An Li, Wenguan Wang, Yixin Zhu, Yizhou Wang

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026

Project, Paper

• Introduce MEDVCR, a clinically grounded counterfactual reasoning framework for medical video diagnosis that models disease-state alternatives with clinical priors.

AdaTracker: Learning Adaptive In-Context Policy for Cross-Embodiment Active Visual Tracking

Kui Wu, Hao Chen, Jinzhu Han, Haijun Liu, Churan Wang, Yizhou Wang, Zhoujun Li, Si Liu, Fangwei Zhong

IEEE Robotics and Automation Letters (RA-L), 2026

Paper, arXiv

• An adaptive in-context policy for cross-embodiment active visual tracking, enabling zero-shot generalization via an Embodiment Context Encoder.

UnrealZoo: Enriching Photo-realistic Virtual Worlds for Embodied AI

Fangwei Zhong* ✉, Kui Wu*, Churan Wang, Hao Chen, Hai Ci, Zhoujun Li, Yizhou Wang

IEEE/CVF International Conference on Computer Vision (ICCV), 2025 (Highlight)

Project, Paper, arXiv, Code

• A rich collection of photo-realistic 3D virtual worlds built on Unreal Engine, designed to reflect the complexity and variability of the open worlds.

VLM Can Be a Good Assistant: Enhancing Embodied Visual Tracking with Self-Improving Vision-Language Models

Kui Wu, Shuhang Xu, Hao Chen, Churan Wang, Zhoujun Li, Yizhou Wang, Fangwei Zhong

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025

Project, Paper, arXiv

• A self-improving framework that enhances Embodied Visual Tracking (EVT) with Vision-Language Models (VLMs) to recover tracking from failure.

Automatic Segmentation and Molecular Subtype Classification of Breast Cancer Using an MRI-based Deep Learning Framework

Xiaoxia Wang, Xiaofei Hu, Churan Wang, Hua Yang, Yan Hu, Xiaosong Lan, Yao Huang, Ying Cao, Lijun Yan, Fandong Zhang, Yizhou Yu, Jiuquan Zhang

Radiology: Imaging Cancer, 2025

Paper

• Develop an MRI-based deep learning framework for automatic breast cancer segmentation and molecular subtype classification.

Clinical Inspired MRI Lesion Segmentation

Lijun Yan, Churan Wang, Fangwei Zhong, Yizhou Wang

IEEE International Symposium on Biomedical Imaging (ISBI), 2025

Paper, arXiv

• Introduce a clinically inspired MRI lesion segmentation framework that leverages multi-sequence MRI priors for accurate and consistent lesion delineation.

Autoregressive Sequence Modeling for 3D Medical Image Representation

Siwen Wang, Churan Wang, Fei Gao, Lixian Su, Fandong Zhang, Yizhou Wang, Yizhou Yu

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2025

Paper, arXiv

• Model 3D medical images as autoregressive sequences to learn general-purpose volumetric representations across organs and imaging modalities.

Empowering Embodied Visual Tracking with Visual Foundation Models and Offline RL

Fangwei Zhong, Kui Wu, Hai Ci, Churan Wang, Hao Chen

The 18th European Conference on Computer Vision (ECCV), 2024

Project, Paper

• Significantly enhanced the understanding and adaptive capabilities of systems in dynamic environments with visual foundation models and offline RL.

Cross-Dimensional Medical Self-Supervised Representation Learning Based on a Pseudo-3D Transformation

Fei Gao, Siwen Wang, Fandong Zhang, Hong-Yu Zhou, Yizhou Wang, Churan Wang✉, Gang Yu✉, Yizhou Yu✉

Medical Image Computing and Computer Assisted Intervention (MICCAI), 2024

Paper

• Integrate 2D and 3D medical imaging data through a pseudo-3D transformation, enhancing the efficiency and effectiveness of SSL for 3D medical image analysis and demonstrating superior performance across various downstream tasks.

Semi-and Weakly-Supervised Learning for Mammogram Mass Segmentation with Limited Annotations

Xinyu Xiong, Churan Wang✉, Wenxue Li, Guanbin Li✉

IEEE International Symposium on Biomedical Imaging (ISBI), 2024

Paper

• Present a semi- and weakly-supervised learning framework for breast mass segmentation that effectively addresses the challenge of identifying small, camouflaged masses in breast cancer diagnosis and the high cost of pixel-wise annotations.

Learning Domain-agnostic Representation for Disease Diagnosis

Churan Wang, Jing Li, Xinwei Sun, Fandong Zhang, Yizhou Yu, Yizhou Wang

The Eleventh International Conference on Learning Representations (ICLR), 2023

Paper

• Disentangle disease-related features from center-effects, enhancing robustness in multi-center image-based diagnosis.

Prediction of Carotid In-Stent Restenosis by Computed Tomography Angiography Carotid Plaque-Based Radiomics

Xiaoqing Cheng, Zheng Dong, Jia Liu, Hongxia Li, Changsheng Zhou, Fandong Zhang, Churan Wang, Zhiqiang Zhang, Guangming Lu

Journal of Clinical Medicine (JCM), 2022

Paper

• Retrospectively analyze clinical imaging data to identify independent predictors of in-stent restenosis following carotid artery stenting, demonstrating that a combination of traditional plaque characteristics and radiomic features provides the highest predictive accuracy for post-procedure outcomes.

Disentangling Disease-related Representation from Obscure for Disease Prediction

Churan Wang, Fei Gao, Fandong Zhang, Fangwei Zhong, Yizhou Yu, Yizhou Wang

International Conference on Machine Learning (ICML), 2022

Paper

• A disentanglement learning strategy under the guidance of alpha blending generation in an encoder-decoder framework (DAB-Net).

Dae-gcn: Identifying Disease-related Features for Disease Prediction

Churan Wang, Xinwei Sun, Fandong Zhang, Yizhou Yu, Yizhou Wang

Medical Image Computing and Computer Assisted Intervention (MICCAI), 2021

Paper

• Learning disease-related representations in medical images by integrating a disentangling mechanism with graph convolutional network, which significantly enhances the performance and interpretability of cancer diagnosis.

Bilateral Asymmetry Guided Counterfactual Generating Network for Mammogram Classification

Churan Wang*, Jing Li*, Fandong Zhang, Xinwei Sun, Hao Dong, Yizhou Yu, Yizhou Wang

IEEE Transactions on Image Processing (TIP), 2021

Paper

• Introduce a counterfactual generative network leveraging the bilateral symmetry prior to enhance mammogram diagnosis performance.

BR-GAN: Bilateral Residual Generating Adversarial Network for Mammogram Classification

Churan Wang, Fandong Zhang, Yizhou Yu, Yizhou Wang

Medical Image Computing and Computer Assisted Intervention (MICCAI), 2020

Paper

• Propose a residual-preserved mechanism with CycleGAN framework based on the bilateral symmetry prior to generate healthy mammogram features for malignancy classification.