“DeepGI” — AI-Assisted GI Endoscopy for Accurate Colorectal Cancer, a Collaboration between Chulalongkorn’s Engineers and Doctor, Shines at ENDO 2024 in South Korea
Faculty members from the Faculty of Engineering and Faculty of Medicine, Chulalongkorn University, jointly developed DeepGI, an AI innovation used in gastrointestinal endoscopy, for accurate cancer detection. The technology is in use at King Chulalongkorn Memorial Hospital and has received global recognition in the GI endoscopy doctors meeting ENDO 2024 in South Korea.
Prof. Dr. Rungsun Rerknimitr, Acting Assistant to the President of Chulalongkorn University for Innovation Affairs, lecturer at the Department of Internal Medicine, Faculty of Medicine, Chulalongkorn University, and a GI endoscopy expert at the Excellence Center for GI Endoscopy, King Chulalongkorn Memorial Hospital, said that the ENDO 2024 in South Korea, organized by World Endoscopy Organization (WEO), in collaboration with the Korean Society of Gastrointestinal Endoscopy (KSGE), saw 3,200 attendants and received live broadcasts from various hospitals so that the participants could witness gastrointestinal endoscopy.
Acting Assistant to the President of Chulalongkorn University for Innovation Affairs
Lecturer at the Department of Internal Medicine, Faculty of Medicine, Chulalongkorn University, and
a GI endoscopy expert at the Excellence Center for GI Endoscopy, King Chulalongkorn Memorial Hospital
The Division of Gastroenterology, Department of Internal Medicine, Faculty of Medicine, Chulalongkorn University, was selected as one in three endoscopy centers around the world to perform a live transmission of endoscopy to ENDO 2024. The transmission showed 4 endoscopy cases, one of which was performed using DeepGI technology, an AI developed by the Faculty of Engineering and the Faculty of Medicine, Chulalongkorn University, to assist doctors in detecting abnormal gastric mucosa before cancer is developed.
Prof. Dr. Rungsun said that the endoscopy using DeepGI technology developed by the Division of Gastroenterology, Department of Internal Medicine, was well received by the audience at the ENDO 2024. Attendees, specifically from India and Italy, have expressed interest in collaborating on research with Chulalongkorn University to further develop this innovation for international applications.
DeepGI — Deep Technology for Gastrointestinal Tracts is developed by Assoc. Prof. Dr. Peerapon Vateekul and Asst. Prof. Dr. Natawut Nupairoj from the Department of Computer Engineering, Faculty of Engineering, Chulalongkorn University, in collaboration with Prof. Dr. Rungsun from the Faculty of Medicine and a research team from the Excellence Center for GI Endoscopy, King Chulalongkorn Memorial Hospital. The technology also received support from E.S.M.Solution Co., Ltd., which is also the distributor of DeepGI. Currently, it is being used at King Chulalongkorn Memorial Hospital and tested at several other hospitals.
At present, the standard method used to detect various abnormalities is endoscopy through the rectum (colonoscopy) or lower gastrointestinal tract. Detecting abnormalities is quite challenging because polyps come in various forms, such as protruding types and flat types that blend with the intestinal wall. They may also be small and have colors that blend with the surrounding area. This makes misdiagnosis easy if doctors lack experience or medical equipment is insufficient. Statistics show that examinations can have an error rate of up to 22 percent.
The DeepGI — Deep Technology for Gastrointestinal Tracts has 4 outstanding features, which are:
- Deep Technology: Its advanced AI technology called Deep Learning is used in real-time detection and characterization of abnormalities with high accuracy.
- Vendor Unlock: It can be used with any brands and models of gastroscopes.
- Extensible Future: It can be further developed in the future to detect abnormalities in other parts of the body, such as the bile duct and the stomach.
- Affordable: It is cheaper than other systems in the market, making it accessible to hospitals or government agencies, which will increase the efficiency of detection and reach provincial and rural doctors.
Assoc. Prof. Dr. Peerapon Vateekul, Department of Computer Engineering, Faculty of Engineering, Chulalongkorn University, explained that currently, DeepGI has expanded its capabilities to detect abnormalities in the stomach, called Gastrointestinal Metaplasia (GIM), which is an early stage of stomach cancer. This type of cancer is one of the common and increasingly occurring cancers, often due to eating habits. This AI-Assisted Solution will help medical staff work more efficiently and reduce errors. DeepGI processes images from videos during endoscopy, then analyzes abnormalities using artificial intelligence (AI) Deep Learning. The model frames areas of abnormality and alerts doctors in real-time, with an accuracy of over 90%, as well as providing accurate characterization of polyp types, whether they are neoplastic or hyperplastic. This capability will increase doctors’ confidence in diagnosing diseases. Research shows that DeepGI can improve polyp detection efficiency by 16%, increase polyp detection by 1%, and can reduce the risk of colorectal cancer by 3%.
Department of Computer Engineering, Faculty of Engineering, Chulalongkorn University
“Colorectal cancer” is one of the most common cancers found in the elderly and is becoming more prevalent in proportion to the growing elderly population. The elderly are at higher risk for gastrointestinal cancers, but early detection of cancer can lead to timely treatment, helping to reduce morbidity and mortality rates from this disease.
DeepGI has been continuously developed and is now at Technology Readiness Level 5 (TRL5). It has been tested in real-world settings in several hospitals and its performance has been benchmarked with competitors in the market. The system can detect abnormalities in both the large intestine and stomach, which will effectively prevent colorectal and stomach cancers. In the future, DeepGI will continue to expand its capabilities, such as detecting abnormalities in the bile duct.
