Project Overview

This project develops an automated system for Gleason grading in prostate cancer histopathology images. Gleason grading is crucial for cancer prognosis and treatment planning, but significant inter-observer variability exists among pathologists. Our AI system aims to provide consistent, objective assessments while reducing diagnostic burden.

Objectives

1. Develop deep learning models that accurately predict Gleason grades from histopathology images
2. Validate the system across multiple institutions and datasets
3. Integrate the system into clinical workflows at partner institutions
4. Create an interpretable system that provides clinically actionable insights

Methodology

• Data: 50,000+ prostate cancer slides with expert annotations
• Architecture: Multi-task learning with grade prediction and tissue classification
• Validation: Cross-institutional validation with prospective testing
• Clinical Integration: Web-based interface for pathologists

Publications from This Project

• Liu, J., et al. (2023). “Automated Gleason Grading in Prostate Cancer.” Journal of Pathology, XX(X), XXX-XXX.

Team Members

• Principal Investigator: Jingxin Liu
• Postdoctoral Researchers: [Names]
• Graduate Students: [Names]
• Clinical Collaborators: [Names]

Current Status

Currently in Phase 2 of clinical validation with prospective testing at partner institutions.