Urgency:
Colorectal cancer ranks 5th among malignant diseases in Vietnam and complete colonoscopy to detect polyp lesions, especially adenomatous polyps and high-risk polyps, is the method that plays the basic role in screening for this disease. The rate of missing colon polyps according to world studies ranges from 20 – 47% and depends on many factors such as the doctor’s experience, time to remove the scope, quality of the mechanical system and colonoscopy preparation. In Vietnam, the big number of patients with digestive diseases and diverse disease patterns put great pressure on endoscopy centers, leading to the risk of quick endoscopy wire removal time and quality of endoscopic images. Colonoscopy can be unreliable and doctors are overloaded and tired. Meanwhile, medical units are currently not qualified to equip endoscopic systems with advanced technology to enhance image quality or tools to support damage detection and identification of cases with malignancy risk. This raises the problem of finding solutions that can both increase the detection rate of colon polyps, especially high-risk cancer groups, while ensuring cost-effectiveness for both medical units and patients.
The goal of the project is to develop effective machine learning algorithms for detecting and localizing colon polyps and classifying lesions at risk of cancer through endoscopic images. From there, a real-time computing device system was built to support endoscopists in detecting colon polyps and diagnosing lesions at risk of cancer.
Social impact:
Technology solutions to support endoscopy doctors through a specialized equipment system for real-time endoscopic image processing will be registered for patents, contributing to affirming the capacity and influence of AI technology application in medical care. Deploying a real-time AI system to support endoscopy doctors at medical facilities, including units with training functions, central and provincial hospitals, will help connect endoscopy doctors, shortening the difference in experience and skills as well as aiming at patients’ rights to enjoy good medical examination and treatment services at not too high costs. The project will also develop a training support platform for endoscopists based on actual images and clinical cases collected at medical units. This will help comprehensively improve, continuously update and standardize knowledge, criteria for evaluating, classifying and identifying colon lesions for endoscopy doctors at many facilities at many levels.
