Making a difference - computer-assisted endoscopic procedures

 

Since 2010, Professor Danail Stoyanov has been developing algorithms to assist endoscopic image and video analysis to support clinicians in routine diagnostic and therapeutic procedures for disease detection in the gastrointestinal tract.

Using artificial intelligence (AI) in the analysis has the potential to save hundreds of lives in the NHS by minimising the numbers of missed lesions in the thousands of diagnostic endoscopic procedures performed annually.

Stoyanov published one of the first papers to use data driven AI algorithms for assisting polyps detection in colonoscopy which significantly improved the robustness of computer assisted detection1 . If detected early during diagnostic screening, polyps can be removed and treatment is curative, preventing colon cancer development. However, polyp detection is highly operator dependent and up to 20% can be missed. Using AI in clinical tools enables complete investigation by helping the endoscopist to find all polyps, ensuring standardised service quality across the NHS.

Professor Laurence Lovat, an endoscopist and specialist in upper and lower gastrointestinal diseases and minimally invasive treatments to prevent and treat gastrointestinal cancer, has pioneered computer assisted real-time endoscopy to support more effective investigations 2,3. In 2019 he was invited to lead the British Gastroenterology Society AI Task Force. AI-assisted endoscopy technology is in weekly use in colonoscopy procedures at UCLH as part of an Innovate UK-funded trial, with promising results, and has recently been applied to oesophaegeal cancer screening.

These algorithms are now a key feature of endoscopic systems using computer assisted image analysis to help detect and diagnose disease, particularly cancer or pre-cancerous growths, and aim to ensure that no polyps are missed during examinations. The technology was used to found Odin Vision Ltd, a company which is developing AI systems for assisting endoscopy as a medical device. Their first product, CADDIE, helps detect bowel cancer by identifying and characterising polyps via live colonoscopy video analysis using AI. Odin Vision was awarded a £1M NIHR AI in Healthcare Award (2020), to enable future installation of CADDIE, a CE marked medical device, in 19 UK hospitals to evaluate its impact on patient outcomes. Odin released CADU in May 2021, the first product for assisting diagnostic investigations of the oesophagus, as a new CE marked medical device powered by AI.

1. Brandao, P et al, J Med Robotics Res, 2018; 2. Ahmad, O et al, Lancet 2019 ; 3. Ahmad, O et al, Tech Gastrointestinal Endoscopy, 2019