M.S. student at Fuzhou University, working on Computational Pathology and Medical Image Analysis.
My research focuses on reliable and label-efficient learning for medical images, especially:
- Fine-grained visual evidence in cytology and pathology
- Weakly supervised & multimodal learning
- Domain generalization & robustness under scanner, center, and dataset shifts
Multi-Granular Text-Guided Weakly Supervised Pathology Image Segmentation
First author · IEEE International Conference on Multimedia & Expo (ICME), 2026
A text-guided weakly supervised pathology segmentation framework that uses multi-granular captions, soft image-text alignment, and collaborative learning to obtain fine-grained segmentation evidence without pixel-level supervision.
Reusable utilities for computational pathology workflows, including patient-level splitting, medical-image evaluation, bootstrap confidence intervals, attention visualization, and WSI inspection.
- Computational pathology: WSI processing, patch construction, cell-level analysis, and slide-level evidence aggregation
- Deep learning: Python, PyTorch, CUDA, Linux; model implementation, training, evaluation, and reproducible experimentation
- Deployment & infrastructure: ONNX, TensorRT, GPU inference, and research computing environments
I am currently exploring reliable medical vision under limited annotation and distribution shifts, with a particular interest in cervical cytology, weak supervision, multimodal learning, and domain generalization.
PhD applicant for Fall 2027 · Open to research discussions and collaborations in computational pathology and medical image analysis.