Researchers use new AI tech to improve polyp detection

Researchers from nine locations across England are using an AI technology to improve polyp detection during colonoscopies allowing doctors to detect signs of bowel cancer at its earliest stages. 

When diagnosed at its earliest stage more than 9 in 10 people with bowel cancer will survive their disease for over five years compared with 1 in 10 when it’s diagnosed late. The study is hoping to recruit over 2000 participants before September 2022.

Colorectal cancer affects 1 in 15 men and 1 in 18 women in the UK with 16,600 deaths every year; it is the UK’s second most deadly cancer. Bowel cancer starts when a polyp (or ‘adenoma’) progresses to cancer, but it can be prevented if detected early enough. Colonoscopy is the ‘gold standard’ assessment for bowel cancer and Adenoma Detection Rate (ADR) (which measures how many polyps the doctor removes) has a notable impact on bowel cancer outcomes. Studies have shown that just a 1% increase in ADR is associated with a 3% decrease in the risk for bowel cancer.

Previous studies (which were limited by being undertaken in very specialist units) have shown that Medtronic’s GI Genius Module could lead to an absolute increase in ADR of 14%. The COLO-DETECT study hopes to support these findings with robust clinical data.

The study lead, consultant gastroenterologist at South Tyneside and Sunderland NHS Foundation Trust and professor of gastroenterology at Newcastle University, Prof. Colin Rees said: “The introduction of AI into colonoscopies is a very exciting breakthrough and we have already seen that it has the potential to significantly increase ADR. Our aim is to assess how much GI Genius can improve ADR when compared to standard colonoscopies without AI. From this study we hope to gather evidence to further the adoption of AI into routine colonoscopy practice across the NHS, helping to reduce bowel cancer deaths across the UK.”

Dr Laura Neilson, consultant gastroenterologist at South Tyneside and Sunderland NHS Foundation Trust, who was the first to recruit a patient into the trial, said: “Bowel cancer is entirely treatable and in most cases curable, especially if it’s diagnosed early - which is why this study is so important. COLO-DETECT is without question, world leading research and we are delighted to be the first hospital in the UK to recruit a patient into this trial and to be leading this pioneering research.”

Michael Anthony Watson, 58, from Sunderland, was invited to take part in the study last November as part of investigations after he started experiencing stomach pains. The COLO-DETECT technology was used during his colonoscopy when it was carried out at South Tyneside District Hospital, where the Trust has set up its trial base, with the AI kit activated as part of the randomised study.

He said: "The clinician asked me if I wanted to take part in this study and I thought if it is going to benefit me, of course, but I knew it would also benefit others, so anything which could help, I was all for. When I went in, they told me the machine was switched on and I wouldn't have known any different. I asked at the end how long it would be before they knew the results and I was told straight away there were no signs of bowel cancer, which was a relief. I know they'd also taken some biopsies, so they were still to check, but at least I knew then what they'd seen. If anyone else was asked to join in the study, I'd say yes, because you also know it's going to help other people."

The GI Genius AI endoscopy module is trained through reviewing more than 13 million polyp images of various shapes and sizes. The algorithm enhances clinicians’ ability to detect polyps during a colonoscopy examination. When it detects a polyp, it alerts the physician with a visual marker on-screen (acting as a ‘virtual’ second observer to the procedure).

Dr Tom Lee, consultant gastroenterologist, has been using the AI technology for four months at Northumbria Healthcare NHS Foundation Trust: “The technology integrates with my hospital’s existing colonoscopy equipment and alerts the endoscopist to the presence of a polyp or other abnormalities with a green box on the screen. It helps us to spot even very subtle polyps – smaller, flatter or partially hidden in a fold of the bowel. I believe that AI could help reduce variation in colonoscopy, among endoscopists but also among individual doctor’s performance. It is well known that doctors tend to spot more polyps or adenomas in the morning than the afternoon for example. When bowel cancer screening colonoscopies are of a consistently high standard, we can hope to improve bowel cancer outcomes across the NHS.”

COLO-DETECT is the first major industry study being run as part of the COLO-SPEED network which is funded by the Sir Bobby Robson Foundation, part of Newcastle Hospital’s Charity. COLO-SPEED aims to radically change how we deliver bowel cancer research – shifting the emphasis to prevention, early diagnosis and screening. We are delighted to have the support of the Sir Bobby Robson Foundation in this important work. 

Sir Bobby’s wife, Lady Elsie Robson said: “When my husband first launched this Foundation, his aim was to attract world-class cancer research to this region, which would then go on to help across the country and beyond. We’re so proud to see this starting with help from COLO-SPEED and I know our charity’s wonderful supporters will join me in wishing everyone involved all the very best. We all want to see research into this disease advancing as quickly as possible.”