3-D Rigid Point Set Registration for Computer-Assisted Orthopedic Surgery (CAOS): A Review From the Algorithmic Perspective

Abstract

Registration is an important problem in computerassisted orthopedic surgery (CAOS). As one typical example, the pre-operative space where the patient-specific surgical plan is usually made needs to be accurately aligned with the patient space where the surgical procedures are conducted. The registration task still attracts a lot of research efforts, partially because establishing the point correspondences between two point sets (PSs) under noise, outliers and partial overlapping is a non-trivial problem. More specifically, in CAOS, (a) the intraoperative points only cover a small partial region of the preoperative whole model; (b) the acquired intra-operative points are usually noisy and contains outliers. To facilitate the related researchers to quickly know both the classical and state-of-theart rigid point set registration (RPSR) methods, we present a concise review about related rigid registration methods in CAOS in this paper. The contributions of this paper include: (1) we review the RPSR methods that are suitable for or very related to the CAOS application; (2) the surveyed registration algorithms’ advantages and disadvantages are discussed and compared; (3) potential future research directions are also discussed.