Deep GI AI: Revolutionizing Gastrointestinal Cancer Detection
The development of Deep GI, an artificial intelligence model, by Chulalongkorn University's Faculty of Medicine and Faculty of Engineering, is poised to transform the detection and diagnosis of gastrointestinal cancers. Trained on hundreds of thousands of endoscopic images, Deep GI aims to match the expertise of senior gastrointestinal specialists in accurately locating colorectal cancer. With plans to spin off into a startup following Thai FDA approval, the technology is expanding to detect and diagnose gastric and bile duct cancers.
Key Features of Deep GI:
- Phase 1: Colorectal Cancer Detection
- Developed in 2018 to address the challenge of limited gastroenterology specialists in Thailand.
- Trained on images from 500 to 1,000 patients, enabling accurate detection of colorectal cancer and polyps.
- Phase 2: Gastric and Bile Duct Cancer Detection
- Trained with new datasets, making it the world's first AI capable of identifying bile duct and gastric cancers.
- Assisting doctors in detecting raised polyps in the colon, and identifying flat lesions in the stomach and small cracks in bile ducts.
- Accuracy and Integration
- Achieves up to 97% accuracy in detecting colorectal cancer, comparable to senior endoscopists.
- Integrates CADx (Computer-Aided Detection and Diagnosis) directly into the system for standard tool use.
- Startup Potential
- Received FDA approval for general patient screening.
- Supported by Chulalongkorn University to expand implementation nationwide through the Ministry of Public Health.
- Private hospitals can adopt the technology through a Chula Startup company.
Impact and Benefits:
- Early Detection: Deep GI aids in early cancer detection, reducing treatment costs and improving patient quality of life.
- Reduced Incidence and Mortality: By expanding screening capacity, the technology aims to lower Thailand's gastrointestinal cancer incidence and mortality rates.
- Scalability: AI-assisted training for doctors, enabling them to perform procedures as effectively as specialists.
- Patient Experience: Preserves the standard endoscopic procedure, minimizing patient preparation and procedure time.
Colorectal Cancer Screening Recommendations:
- Prof. Dr. Rungsun recommends regular screening for individuals aged 50 and above, with earlier screening for those with a family history of colorectal, gastric, or breast cancer.
- Colonoscopy offers advantages over other screening methods, allowing for immediate removal of suspicious lesions and reducing procedure time and risks.
