Eye-tracking fluoroscopy system: a new artificial intelligence-based system to communicate activate watching of the monitor during endoscopy.

Minimizing radiation exposure time is one of the three principles of radiation safety. However, fluoroscopy is often used unnecessarily when endoscopists are not looking at the fluoroscopy monitor. Unnecessary radiation exposure is due to the complexity of fluoroscopic endoscopy procedures, wherein multiple operators independently view different parts of multiple monitors. These procedures usually need an endoscopist for endoscope maneuvers and one or more assistants for other devices; therefore, a radiologic technologist, who manages radiation exposure, cannot always monitor the line of sight of all operators at once. As limited tools are available to reduce endoscopic radiation exposure time [1], we developed a new eye-tracking fluoroscopy system that informs the radiologic technologist when the operators’ line of sight is directed toward the monitor. It can track the operators’ line of sight and flash a “watching” alert on the detection monitor once a line of sight to the fluoroscopy monitor is detected. This system, jointly developed with NEC Corporation (Tokyo, Japan), was designed to detect remote gaze of operators simultaneously using artificial intelligence (AI)-based face-recognition technology. Radiologic technologists can use this system by attaching an external camera to the fluoroscopy monitor ([Fig. 1]) and connecting the camera to the detection monitor and the AI software.