A team from Yokohama University in Japan have developed a new computer-aided endoscopic and diagnostic system.
The system was trained using 30,000 images of colorectal polyps and checks around 300 features of the polyp, which it does in less than a second. The results were encouraging, with the AI being able to fairly accurately identify which polyps were likely to be cancerous in realtime. The AI will require further training, but is now set to enter clinical trials. The team is also working on an automatic polyp detection system.