AI-Assisted GI Endoscopy Enables Accurate Detection of Colorectal Cancer

By HospiMedica International staff writers
Posted on 19 Aug 2024

Image: The gastrointestinal disorder detection system (DeepGI) procedure (Photo courtesy of Chulalongkorn University)

Colorectal cancer is among the most prevalent cancers affecting the elderly and is increasing in frequency alongside the growth of the elderly population. Early detection can facilitate timely interventions, significantly reducing morbidity and mortality associated with this disease. Currently, colonoscopies and lower gastrointestinal endoscopies are the standard methods for detecting abnormalities. However, the identification of such abnormalities is challenging due to the diverse forms of polyps, including protruding and flat types that blend into the intestinal wall. These polyps may be small and similarly colored to their surroundings, which increases the likelihood of misdiagnosis, particularly when doctors lack experience or when there is insufficient medical equipment. Statistics indicate that diagnostic error rates in such examinations can be as high as 22%. Now, a novel device designed for the rapid detection of gastrointestinal cancer delivers accurate results, enhancing preventive medicine in gastrointestinal malignancy and reducing the incidence of cancer.

The DeepGI, or Deep Technology for Gastrointestinal Tracts, developed by Chulalongkorn University (Bangkok, Thailand), is an advanced artificial intelligence (AI) tool intended to enhance the detection of abnormal polyps in the colon. Utilizing deep learning technology, DeepGI performs real-time detection and characterization of abnormalities with high accuracy, thereby augmenting the diagnostic capabilities of endoscopists. The system is compatible with various endoscopic procedures, including colonoscopy, gastroscopy, and cholangioscopy. Designed specifically for colonoscopy, DeepGI boasts a polyp detection accuracy of over 95%. In gastroscopy, it can delineate areas of Gastrointestinal Metaplasia (GIM) in the stomach, and for cholangioscopy, it can classify benign areas in real time. Thus, DeepGI offers a comprehensive solution for the entire gastrointestinal tract. Beyond mere detection, DeepGI can also differentiate between malignant (neoplastic) and benign (hyperplastic) tissues, potentially obviating the need for biopsies.

DeepGI supports a wide array of endoscopy cameras, regardless of vendor or model, facilitating its integration into existing operational setups in medical facilities. It has been validated by seasoned endoscopists who affirm its efficacy in enhancing anomaly detection rates. Currently, the technology is operational at King Chulalongkorn Memorial Hospital and is undergoing testing at several other hospitals. It has garnered international acclaim at the GI endoscopy doctors meeting, ENDO 2024, in South Korea. DeepGI holds significant promise for broader application, particularly as a support tool for endoscopists in community hospitals, where there is often a lack of medical and technological resources.

Related Links:
Chulalongkorn University