Tracking patellar osteophytes to detect osteoarthritis

ABSTRACTThe presence of bone spurs (osteophytes) in knee-joints provides an indication of the onset of osteoarthritis and related deformities in patients. The shape of the patella bone at knee-joint undergoes various changes with the advancement of the disease, which may be discerned accurately from X-ray images. The main objective of this work is to identify the presence of osteophytes in the patella bone of knee-joint. Geometric features are extracted that highlight protrusions present on the bone surface from lateral views of X-ray images. The patella region is extracted from the image by performing texture-based segmentation followed by adaptive thresholding. Characteristic features are extracted from the segmented patella using derived novel measures based on differential chain code of the contour and convex hull. Linear support vector machine is used to classify diseased test-cases from healthy ones using the extracted features. Experimental results demonstrate the capability of the proposed image-analysis tool for detecting osteoarthritis with average accuracy of 93.75%.