A decision support system for the automatic assessment of hip osteoarthritis severity by hip joint space contour spectral analysis

Abstract

A decision support system was developed for the grading of hip osteoarthritis (OA) severity. Sixty four hips (18 normal, 46 osteoarthritic) were studied from the digitized radiographs of 32 patients with unilateral or bilateral hip-OA. Hips were allocated into three OA-severity categories, formed accordingly to the Kellgren and Lawrence scale: "Normal", "Mild-Moderate", and "Severe". Employing custom developed algorithms: (i) the radiographic contrast was enhanced, (ii) 64 ROIs, corresponding to patients' radiographic Hip Joint Spaces (HJSs), were determined, and (iii) Fourier descriptors of the HJS-ROIs boundary were generated. These descriptors were used in the design of a two-level hierarchical decision tree structure, employed for the discrimination of the OA-severity categories. The overall classification accuracies accomplished by the system, regarding the discrimination between: (i) Normal and osteoarthritic hips, and (ii) hips of "Mild-Moderate" OA and of "Severe" OA were 92.2% and 86.0%, respectively. The proposed system may contribute to osteoarthritic patients management. © Springer-Verlag Berlin Heidelberg 2006.