Computer Vision News - March 2023

30 Bringing CV and Robotics to the OR Total knee arthroplasty is a widespread and rapidly growing procedure. As a standard manual operation, it is invasive and associated with unsatisfactory and inconsistent patient outcomes. Even technological solutions offered by other MedTech companies are less than ideal. “ Surgeons tell us it’s like woodworking! ” Marion laughs. “ They use a hammer, a screwdriver, and mechanical tools to localize the patient during the procedure. Our competitors screw percutaneous 3D trackers on the patients. They’re like stars with balls at the end of their branches, tracked by a localizer. That’s a big robot with a big footprint on the OR. You have to hurt the patient more than in the standard procedure. ” As part of the procedure, the femur and tibia must have several plane cuts. These are related to the good placement of the implant and better patient outcomes, so they must be exact. In Ganymed’s solution, a computer vision algorithm and a vision sensor embedded in a robot help localize the patient intraoperatively to precisely guide the surgeon’s gesture. It is minimally invasive , avoiding attaching extra trackers tothepatient’sboneormakingunnecessary cuts. “ We want a really intuitive and compact robot, ” Marion tells us. “ My team wants robust algorithms that work for different operating room conditions and patient morphologies. This procedure has to work for all patients! ” The core innovation of the vision part of the system involves a novel approach to intraoperatively locating a patient’s bone using the embedded vision sensor, which collects data during the operation. Ganymed Robotics is a robotics start-up founded in 2018 aiming to improve the surgical practice in orthopedics by bringing computer vision algorithms and robotics to the operating room. Marion Decrouez, Computer Vision Team Lead at Ganymed, speaks to us about its work developing a robot for total knee arthroplasty.