Measurement of Acetabular Wear Using Intelligent Ellipses

A summary of recent studies reported that of a sample of 100 Total Hip Replacements (THRs) 18 revisions occurred, 16 of which were due to osteolysis and loosening of the acetabular cup. Thus tracking acetabular wear is a key step in the early detection of failure in THRs. The femoral head and rim of the acetabular cup appear elliptical in radiographs regardless of orientation. This knowledge of shape can be used to measure acetabular wear. Current approaches range from manual methods that estimate the centre of the femoral head using transparent overlays of concentric circles to semi-automated methods using Sobel edge detection. Our aim is to create an automated, accurate and repeatable method that improves on existing performance. It involves the use of intelligent ellipses that learn to identify the typical change of the grey level at the femoral head from a set of manually annotated digitized radiographs. When trained these intelligent ellipses are placed within a previously unseen digitised radiograph of a THR. Searching occurs along normals to the intelligent ellipses and comparing the normalized grey level derivative profiles (a measurement of the gradient) to a statistical model of the typical grey level derivative around a femoral head. The location and other parameters of the intelligent ellipses are altered to best match the model. After several searches the ellipse converges around the femoral head in the radiograph. In a similar fashion the rim of the acetabular cup can be located. Once the centres have been found the distance from the centre of the femoral ellipse to the centre of the acetabular ellipse can be calculated.