Artificial Intelligence in Colonoscopy
Artificial Intelligence (AI) is already in use in gastrointestinal endoscopy (GIE), as well as many other areas of clinical medicine and healthcare. Its use in GIE, just like in other areas is expanding rapidly and will become the new standard of care in the very near future. Here, I discuss the main current application of AI in GIE – computer-aided [polyp] detection and characterisation – as well as explore the main emerging and near-future applications. This presentation primariily takes the perspective of UK practice, although it is broadly applicable globally also. Computer-aided detection (CADe) is largely (but not universally) supported by the current evidence base. Further trials and systematic reviews are underway to further develop and refine our understanding of how to optimally apply CADe to GIE. Following hot on the heels of CADe is polyp characterisation, or computer-aided diagnosis (CADx); working in combination, CADe and CADx together locate and provide an optical diagnosis of polyps to support endoscopists in deciding what to do with them. Here is where the nuances of applying this technology really begin to emerge. Finally, other applications of AI in GIE which are in existence but not yet in common use include: procedure duration recording, caecum and appendiceal orifice recognition, bowel preparation quality scoring, mucosal visualisation assessment, instrument recognition, and tissue recognition to support advanced polypectomy. With these technologies in place it is only a small step to automated reporting and real-time quality control and improvement, which could revolutionise the training and accreditation of endoscopists and endoscopy services.
