The team designed a prototype for the Ear, Nose, and Throat Research & Development division at Medtronic to improve CT scan registration prior to surgery with sub-millimeter accuracy. Moving away from traditional 3D scan tracing methods used in Medtronic’s Stealth Station, the prototype aimed to simplify and accelerate the registration process by employing computer vision. It integrated Perspective-n-Point and stereoscopic vision algorithms into a handheld, surgically-safe device, enabling efficient tracking of patient facial movements during surgical procedures.
The team developed an Initial Guess algorithm that combined 2D images and 3D CT scan coordinates to approximate the patient’s facial position relative to the camera system and electromagnetic tracker. The prototype successfully fulfilled all performance requirements, achieving an accuracy of 0.6 millimeters. These results demonstrate the effectiveness of the approach and highlight its potential to significantly advance the accuracy and efficiency of surgical registration in future applications.
