|
Getting your Trinity Audio player ready...
|
The Faculty of Medicine at the Chinese University of Hong Kong (CU Medicine) conducted two studies on the application of artificial intelligence (AI) in endoscopic technology, showcasing its potential to significantly improve the diagnosis and treatment of gastrointestinal (GI) cancer.
These studies, published in prestigious international journals such as Nature Communications and Clinical Gastroenterology and Hepatology, mark a significant step forward in the integration of AI into medical practices.
Professor Philip Chiu Wai-yan, who serves as the Professor and Head of Upper Gastrointestinal and Metabolic Surgery in the Department of Surgery at CU Medicine, as well as the Director of the Chow Yuk Ho Technology Centre for Innovative Medicine, emphasised the alignment of these studies with the goals outlined by the National Health Commission of the People’s Republic of China (PRC).
The commission has advocated for the promotion of minimally invasive surgery, including endoscopic interventions, as part of the National 14th Five-Year Plan. Professor Chiu believes that CU Medicine’s expertise in healthcare and innovation positions the institution to contribute significantly to national healthcare goals.
The first study focused on AI-assisted colonoscopy for the detection of colorectal cancer. Colorectal cancers often originate from colorectal polyps, with adenomas being the most prevalent type. To effectively prevent colorectal cancer, identifying and removing adenomas during colonoscopies is crucial.
However, the accuracy of adenoma detection is influenced by various factors, including the size, location, and morphology of adenomas, as well as the experience of the endoscopist. CU Medicine introduced AI-assisted colonoscopy in early 2021, enabling real-time analysis of endoscopic images to alert doctors to the presence of adenomas and tumours.
The research team at CU Medicine conducted a study between April 2021 and July 2022, involving 22 junior endoscopists-in-training with less than three years of experience and fewer than 500 endoscopies performed. The results were compelling, demonstrating that these less experienced endoscopists achieved a remarkable 40% increase in adenoma detection rates with the assistance of AI.
This improvement was even more pronounced among those at the beginner level. The study underscores the potential of AI to enhance the skills of less experienced medical professionals, particularly during their initial learning phases.
Dr Louis Lau Ho-shing, an Assistant Professor in the Department of Medicine and Therapeutics at CU Medicine, highlighted the significance of the research for the future development of AI in clinical medicine and endoscopic training. He stressed that junior endoscopists, often less skilled and requiring higher levels of assistance during their early training, can benefit significantly from AI, providing standardised imagery guidance to enhance their skills.
The second study conducted by CU Medicine focused on developing an AI surgical assistance platform named AI-Endo. This platform is designed to aid in endoscopic submucosal dissection (ESD), a therapeutic procedure introduced by CU Medicine in Hong Kong in 2004 to resect early-stage GI cancer. ESD is a complex procedure that demands specialised technical skills and experience from endoscopists. With an increasing number of early-stage GI cancer detections through population screening programs and diagnostic AI systems, there is a growing need to disseminate ESD techniques.
To train AI-Endo, the research team compiled 47 complete ESD video clips spanning 12 years, consisting of over 2 million frames. This extensive dataset allowed the AI system to learn the entire surgical workflow and analyse real-time actions during ESD. Professor Dou Qi, an Assistant Professor in the Department of Computer Science and Engineering at CUHK, explained that AI-Endo can discern real-time procedures, guide doctors through the process, and provide data for postoperative analysis and review.
Dr. Yip Hon-chi, an Assistant Professor in the Department of Surgery at CU Medicine, emphasised the time and procedural competency required for endoscopists to perform ESD independently. AI-Endo, by shortening the training time and providing a reliable tool, offers a valuable solution to assist endoscopists in mastering the ESD procedure efficiently.
