ANATOMIC REGISTRATION FOR ARTHROPLASTY USING AN ACTIVE CONSTRAINT ROBOT

Abstract

Introduction: Active Robots have been shown to be effective at performing arthroplasty, but some hesitation has been felt by the surgical world. The lack of human interface in the procedure has been one of the stumbling blocks towards wider acceptance. The Acrobot has been developed, at Imperial College London, in collaboration with University College London to allow the surgeon to perform the surgery himself, but with active constraint, preventing him from taking too much bone, or straying into soft tissue.

Materials and methods: A preoperative planning system is used, based on ct data acquired without fiducial markers. Semi-automated segmentation is performed. The surgeon then performs the virtual surgery on the bones on screen, allowing precise sizing, and orientation. The safe field of activity is then defined, within which the surgeon is free. The patient is positioned on the operating table and immobilised. Anatomic registration is then performed, and when sufficient accuracy obtained, the milling procedure is begun. A high speed electric milling tool is used, and with it the bone planes are prepared sequentially. The prosthesis is then inserted in standard fashion.

Results: Laboratory testing on dry bone and cadaveric models have confirmed that the registration process is now accurate. At the moment we are using a classical ICP algorithm to register the data points. For this test the Root Mean Square is 0.626 mm in a cadaveric model. This pinless anatomic registration can be achieved rapidly, if the initial siting points are accurately identified.

Conclusion: The active constraint concept seems to be a safe and user friendly way of achieving robotic level accuracy with a human touch. Anatomic registration using the robot is accurate, and early clincal trials of total knee arthroplasty are encouraging.