Research Directions
My research focuses on medical artificial intelligence, with expertise spanning the entire pipeline from mathematical modeling theory to open-source algorithm software, clinical applications, and industrial transformation.
Medical Image Computing and Analysis
- Interactive 3D Medical Image Segmentation: Developed the Grow From Seed interactive segmentation algorithm, which has become a core module in 3D Slicer, the world's largest open-source medical image computing software platform.
- Statistical Shape Analysis: Developed novel shape analysis methods based on 3D Poisson equation with quantitative validation framework.
- Multi-scale Shape Representation: Developed wavelet-based shape decomposition and reconstruction techniques for 3D shapes of arbitrary topology.
- Medical Image Registration: Developed algorithms for deformable image registration using diffusion models and Mamba-based networks.
AI for Medical Diagnosis and Screening
- Breast Cancer Screening: Developed AI-assisted automated breast ultrasound systems for large-scale screening, deployed in over 70 medical institutions serving more than 90,000 people.
- Computational Pathology: Developed deep learning models for digital pathology image analysis, including nucleus segmentation, gland segmentation, and cancer classification.
- Radiomics and Predictive Modeling: Developed radiomic analysis pipelines for predicting treatment response and prognosis in various cancers.
- Medical Image Synthesis: Developed diffusion probabilistic models for synthesizing medical images with annotations to improve segmentation performance.
Open Source Software and Algorithm Development
- 3D Slicer Integration: Multiple algorithms integrated into 3D Slicer, including Robust Statistics Segmenter and Grow From Seed interactive segmentation.
- Insight Toolkit (ITK) Contributions: Spherical wavelet ITK filter and Conformal flattening ITK filter.
- Open Source Platforms: Developed computational histopathology platform and interactive 3D medical image segmentation solutions.
Clinical Translation and Industrial Applications
- Medical Device Certification: Developed "3D Breast Ultrasound Image Management Software" which obtained Class II Medical Device Registration Certificate - the first of its kind in China.
- Clinical Impact: Increased breast cancer detection rate from 0.5‰ to 3.13‰ through AI-assisted screening technology.