Submission to CAOS 2014 Computer-assisted Total Hip Arthroplasty: from Pre-operative Planning to Post-operative Assessment J. Schmid 1 , C. Chênes 1 , S. Chagué 2 , P. Hoffmeyer 3 , P. Christofilopoulos 3 , M. Bernardoni 4 , C. Charbonnier 2 1 Geneva Health School (HEdS), HES-SO, Geneva, Switzerland 2 Medical Research Department, Artanim Foundation, Geneva, Switzerland 3 Orthopedics and Trauma Service, University Hospitals of Geneva, Geneva, Switzerland 4 Medacta International SA, Lugano, Switzerland Introduction Total Hip Arthroplasty (THA) surgery restores the articulation mobility and stability by implanting a prosthetic hip joint. The selection and positioning of the implant are critical in THA. Conventional planning seeks the optimal selection of implant characteristics by relying on a “static” approach which only exploits anatomical cues derived from a plain radiograph or a CT scan. Dynamic aspects are indeed neglected, such as pelvic tilt during postural changes and the necessary motion of the prosthetic hip to yield a satisfactory range of motion (ROM) in everyday life. Despite an adequate planning is produced, its execution during the surgical act remains problematic, since intra-operative guidance is not commonly provided. As a result, the effective positioning of implants may significantly deviate from the planning. A post-operative assessment of the prosthetic hip is hence necessary to investigate the quality of the surgery. We present our computer-assisted framework “MyHip” for THA. The framework provides intra- operative assistance based on personalized guiding blocks. However, we will not cover in this paper this intra-operative support; instead we will focus on the pre-operative planning and post-operative analysis. Methods Pre-operative planning In the pre-operative stage, we reconstruct the bones of the patient’s hip joint from a CT image. We usually adopt a semi-supervised segmentation. First, an automatic segmentation is applied (e.g., intensity thresholding or a more advanced method like our approach based on physically-based deformable models [1]). Second, we perform possible manual corrections to refine areas of interest (e.g., acetabulum area) and correctly tackle pathological areas. Pathological areas are recurrent in operated hips and they affect the accuracy of automatic approaches (deviation from “normal” intensities and morphology).