Digital Pathology AI Pipeline
End-to-end WSI and multimodal analysis workflows for histopathology
This project focuses on building end-to-end artificial intelligence pipelines for digital pathology, including whole-slide image preprocessing, tile extraction, quality control, tissue classification, segmentation, annotation workflows, and multimodal pathology analysis.
The work is part of my postdoctoral research at Stanford University in the Department of Pathology and the Center for AI in Medicine & Imaging.
Role: Postdoctoral researcher
Key contributions:
- Built end-to-end pipelines for whole-slide image and tile-level tissue classification and segmentation.
- Implemented robust preprocessing workflows for artifact filtering, tile extraction, and quality control.
- Trained and deployed pathology foundation models using image and text inputs for multimodal analysis and reporting.
- Annotated tumor regions in whole-slide images using QuPath.
- Created high-quality ground truth and quality assurance workflows for model development.
- Curated large-scale tile-level tissue datasets for benchmarking and reproducible computational pathology research.
Research focus:
- Whole-slide image analysis
- Computational pathology
- Medical AI
- Foundation models
- Multimodal pathology analysis
- Tissue classification and segmentation