INTRA-OPERATIVE IMAGE REGISTRATION FOR MINIMALLY INVASIVE SURGERY

Abstract

A major limiting factor for the accuracy in Computer Assisted Surgery (CAS) is the system’s positional knowledge of the patient’s anatomy, derived through the process of registration. In computer assisted Minimally Invasive Surgery (MIS) the registration process is made more difficult by the lack of direct access to a large portion of the surface to be registered. Current experience with a hands-on robotic surgery system, which uses a set of points measured with a mechanical digitiser on the exposed surface of the bone and a surface reconstructed from computer tomography (CT) data, has shown that accurate and robust registration is still possible through an MIS approach.

The registration method described here, which was originally developed for robotic assisted total knee arthroplasty (TKA), has successfully been adapted for robotic assisted unicompartmental knee arthroplasty (UKA) and computer assisted hip resurfacing (HR). Results show that good registration can be achieved by registering the bone surfaces through conventional surgical incisions, with two additional stab-wounds required for the UKA procedure. However, experimental results suggest that, because of the limited access resulting from a smaller incision, a good correspondence between the point-set and surface measurements (i.e., better than one millimeter) is necessary for registration accuracy better than two degrees and two millimeters. This degree of correspondence can be expected for a good surface model and an appropriate intra-operative setup, but poses an important constraint on the requirements for a system suitable for this type of procedure, if a registration method based on anatomical features is to be used without the need for additional access.