I build foundation models and clinical AI systems for pathology, under the supervision of Dr. Hao Chen in the HKUST SMARTX Lab. Before HKUST I studied at UESTC with Prof. Liang-Jian Deng, working on remote sensing image fusion.
01Research
My work sits between Healthcare AI and Computer Vision: pathology foundation models, whole-slide image analysis, and the systems that put them in front of clinicians. Hover a project to preview it.
Healthcare AI · Pathology
PMIL Shapley-value MILHMIL Hierarchical MILPathLUPI Genomic distillationPathBench PFM benchmark
Computer Vision · Foundation Models
GameGen-X Game video generationLPPN Laplacian pyramidFusionNet Multi-scale fusion
02Selected Publications
All selected →First and co-first author work. For the complete list see Google Scholar.
Co-first author · Cover article
DeepPlaque: a scalable multimodal platform for Aβ pathology and cell analysis in Alzheimer’s disease
EMBO Molecular Medicine, 2026
First author
Genome-Anchored Foundation Model Embeddings Improve Molecular Prediction from Histology Images
Nature Communications, 2026
Co-first author · Preprint
Generalizable Cervical Cancer Screening via Large-scale Pretraining and Test-Time Adaptation
arXiv preprint, 2025
First author
Learning with Less Supervision: A Survey of Label-efficient Learning for Medical Image Analysis
Medical Image Analysis, 2026
First author
HMIL: Hierarchical Multi-Instance Learning for Fine-Grained Whole Slide Image Classification
IEEE Transactions on Medical Imaging, 2024
First author
Laplacian Pyramid Networks: A New Approach for Multispectral Pansharpening
Information Fusion, 2022
Selected collaborations
- A Multimodal Knowledge-enhanced Whole-slide Pathology Foundation Model (mSTAR)Nature Communications, 2025
- Towards A Generalizable Pathology Foundation Model via Unified Knowledge Distillation (GPFM)Nature Biomedical Engineering, 2025
- GameGen-X: Interactive Open-world Game Video GenerationICLR 2025
- PathBench: A comprehensive comparison benchmark for pathology foundation modelsarXiv, 2025
- Detail Injection-Based Deep Convolutional Neural Networks for Pansharpening (FusionNet)IEEE TGRS, 2020
03Platforms
Beyond papers, I design and code the software that carries this research into practice. All three platforms below were built from scratch.
Built from scratchLogin required
SmartCyto
A web platform for AI-assisted cytology screening, bringing whole-slide image management, analysis workflows, and task tracking into one workspace.
Built from scratchLogin required
SmartPath
Brings foundation-model-based pathology analysis into an integrated web workflow, connecting slide management, AI inference, visualization, and structured reporting.
Built from scratchLogin required
PathLab
A workspace for computational pathology research: whole-slide exploration, dataset organization, and AI-assisted experimentation. Its research agent connects research questions with experiment planning, execution, and evidence-linked reports.








