AI learned how to suture wounds by watching videos from surgical operations

Researchers from the University of Berkeley, Intel and Google Brain taught an AI model to operate by simulating video recordings

eight surgeons at work. The algorithm, called Motion2Vec, was trained on footage where doctors control surgical robots to apply sutures or tie knots. But while the robot is usually controlled by a doctor from a computer console, in the case of Motion2Vec it does it independently.

He has already shown his skills in stitching the piecesfabrics. In tests, the system reproduced the surgeons' movements with an accuracy of 85.5%. Achieving this level of accuracy was not easy: the eight surgeons in the videos used a variety of techniques, so the AI ​​had to choose the best option.

To solve this problem, the team usedsemi-autonomous algorithms that study a problem by analyzing partially labeled data sets. This allowed AI to understand the basic movements of surgeons from a small amount of data. However, researchers acknowledge that the system needs to be refined before engaging in operations on its own.

Now scientists plan to conduct tests withdifferent types of tissues so that the system can adapt to different situations - for example, to unexpected bleeding. The next step in the development of the system will be semi-automatic remote surgery. At this stage, the robot will assist the doctor.

Scientists want to create more AI using the samemodels. There is a lot of unstructured information on the Internet  in the form of videos, images and text. Robots can extract useful content from it to make sense of the data and help us solve everyday problems.

Read also

— Look at which organs the coronavirus attacks first and how it happens

— Who was the first organism on Earth? Forget everything you knew about the origin of life

— Check out photos of the eclipse taken around the world