Laparoscopic surgeries are often automatically recorded from the point of view of the endoscope’s lens. This is thanks to built-in recording equipment that accompanies many commercial endoscopic systems. What isn’t easy is reviewing all those hours of footage to find something that may be useful for training clinicians or that may be used to improve laparoscopy-related equipment.
Now researchers at MIT have reported at the International Conference on Robotics and Automation in Singapore on a new video processing system that can, on its own, identify different stages of laparoscopic surgeries, potentially allowing researchers to quickly find relevant scenes that they can easily study.
The computer vision algorithm powering the system can spot when a biopsy is performed, a wound irrigated, or tissue stapled, among other activities. Other actions can also be found among recordings.