NEC Corporation has developed “WISE VISION Endoscopy,” an AI diagnosis-support medical device software for colonoscopies. The company began selling the software in Japan from January 12. It will also begin selling it in Europe in the near future. The software connects with existing endoscopy equipment, and AI is used to automatically mark potential lesions from images taken during endoscopic procedures, to support the detection of lesions by doctors during the procedures.
Colorectal cancer is said to be the most common cancer in Japan and the second most common in Europe also. Colorectal cancer is known to originate from precancerous lesions (colorectal neoplastic polyps), and it is possible to suppress the progression to cancer by detecting and removing lesions at the polyp stage during endoscopy procedures.
However, polyps need to be found with the naked eye of the endoscopist, and are often difficult to detect because of their size and difficult-to-recognize shape, causing approximately 24% to be missed and thereby delaying treatment.
In light of that, NEC has been working with the National Cancer Center Japan since 2016 in order to resolve this issue, and has developed software that can automatically mark potential lesions based on the use of AI to learn from endoscopic images of more than 10,000 lesions, as well as learning from the observations of expert physicians. This software was developed by applying NEC's AI technologies—“NEC the WISE”—and its face recognition technology, which is highly evaluated by the National Institute of Standards and Technology in the U.S.
This software can be connected to the endoscopic systems of three leading manufacturers. By simply connecting an existing endoscope to a monitor and terminal equipped with the software, users can start using it immediately. It is easy to prepare and can also be moved within hospitals. It can therefore be used efficiently wherever an examination is taking place.
Additionally, the system is notable for its highly visible and flexible interface, and it makes the lesion candidates known with a notification sound and marking. The notification sounds, volume, and marker colors can be customized at any time according to the user's preferences. Thanks to its high-visibility user interface the system can be operated intuitively, allowing users to smoothly proceed with examinations.