Total hip replacement in acute femoral neck fractures is done in salvage situations such as non-union and aseptic necrosis. Loosening of the prosthesis from the host bone has been recognised as a source of the most common post-operative complications. Therefore, continued follow up is necessary to identify early evidence of migration of the prosthesis. In current practice, clinicians assess loosening by visually inspecting a number of standard x-ray images of the patient's hip joint, taken over a period of time. This process is time-consuming and introduces inter-observer and intra-observer variability. The present thesis provides a new algorithm for computer-aided evaluation of prosthesis loosening in the form of an ???expert assistant???. / Computer-aided measurement of prosthesis migration involves finding a discrete mapping between features of the prosthesis model and features in the X-ray image. The orientation and the position of patient relative to the camera do not remain fixed in each radiograph taken over a long period. As a preliminary to this analysis, it is necessary to search for comparable radiographs in which the poses are similar. / This thesis introduces a robust geometrical matching algorithm to describe the pose of the prosthesis in a radiograph and the location of its key feature points relative to femur. An algorithm that uses the expectation maximisation (EM) approach has been designed and implemented to estimate the parameters of the rotation matrix and the translation vector for the 3D reference model of the prosthesis. This is achieved by matching selected control points on the reference model with corresponding set of control points extracted from X-ray image without the knowledge of their correspondence match. For matching purposes, feature points are extracted from X-ray images using novel and existing image processing techniques. Once they are located, the parameters of migration are measured. A Kalman filter-based bone edge detection algorithm is proposed and successfully implemented in this thesis. / Thesis (PhD)--University of South Australia, 2001.
| Identifer | oai:union.ndltd.org:ADTP/285379 |
| Creators | Abeynayake, H T G Canicious M |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | copyright under review |
Page generated in 0.0019 seconds