Researchers at the Chinese University of Hong Kong’s medical faculty have discovered that artificial intelligence (AI) can greatly assist junior doctors in identifying colon tumours more accurately during colonoscopies. The AI technology, acting as an extra pair of eyes, proved particularly effective at detecting smaller growths known as adenomas that might otherwise go unnoticed. This development is promising in reducing the miss rate for spotting potential cancerous lesions, according to Dr Louis Lau Ho-shing, an assistant professor of medicine and therapeutics.
Traditional colonoscopies require a high level of multitasking, leaving doctors who are fatigued or less experienced susceptible to missing smaller lesions. Dr Lau emphasized the importance of not overlooking these lesions, as they have the potential to develop into cancer in the future. By using AI tools to identify potential tumours in video footage obtained during colonoscopies, doctors can improve their detection rates and ensure early intervention.
Colorectal cancer is both the second most prevalent form of the disease and the second-highest cause of all cancer-related deaths in Hong Kong, following lung cancer. In 2021, the city reported 5,899 new cases, comprising 15.3% of all newly reported occurrences of the disease. To address this issue, the government has implemented a subsidized screening program for residents aged between 50 and 75.
The Chinese University of Hong Kong’s medical faculty, which has been exploring advanced technology in cancer treatment for nearly two decades, began utilizing AI in colonoscopies in 2021. While previous studies have shown AI’s ability to improve tumour detection rates in experienced doctors, this research focused on its potential benefits for less experienced doctors.
Between April 2021 and July 2022, the research team recruited 22 junior doctors, each having performed fewer than 500 colonoscopies, to conduct the procedure on 766 patients. Out of these, 386 doctors utilized the AI tool, while the remaining doctors followed conventional methods. The published results indicated that the group using AI achieved an overall adenoma detection rate of 57.5%, compared to 44.5% for the other group. The AI tool particularly excelled in detecting small tumours with a diameter under 5mm, with a detection rate of approximately 40% for doctors using AI, compared to 25% for those without AI. For medium-sized tumours, the detection rates stood at around 36% and 29% respectively.
Notably, less experienced doctors who had conducted fewer than 200 colonoscopies benefited significantly from using AI, achieving a 60% detection rate, compared to approximately 42% for their counterparts. While a few public hospitals in Hong Kong have already adopted the AI detection tool, no guidelines have been issued regarding its widespread application. Dr Lau hopes that these findings will encourage the tool’s broader adoption to improve patient outcomes.
In addition to improving colonoscopy procedures, the Chinese University of Hong Kong’s medical faculty has developed an AI-based platform called AI-Endo to aid in training doctors to perform early-stage gastrointestinal cancer surgery. This challenging procedure allows patients with stage one tumours to undergo non-invasive removal, although it demands a high level of skill and experience.
Traditionally, doctors required almost 100 operations to attain competence in this procedure. However, AI-Endo aims to shorten the learning curve. Built over two years with the collaboration of the medical and computer science and engineering departments, AI-Endo utilizes twelve years’ worth of surgical data, potentially comprising one of the world’s largest datasets for this purpose. The tool enables doctors to accurately predict the necessary steps in real-time during surgery and generates post-operative reports.
Dr Yip Hon-chi, an assistant professor of surgery, expects to initiate clinical trials for AI-Endo next year. This development has the potential to revolutionize the field by reducing the time and number of operations required for less experienced doctors to master complex surgical procedures.
The application of artificial intelligence in colonoscopies and early-stage gastrointestinal cancer surgery shows promise in enhancing the accuracy and effectiveness of these procedures. With further research and adoption, AI has the potential to improve patient outcomes, reduce healthcare costs, and contribute to the fight against cancer.
