A novel hybrid of S2DPCA and SVM for knee osteoarthritis classification

Abstract

A computer-based system was designed for grading and quantifying knee osteoarthritis (OA) severity. This paper presents a novel approach to knee osteoarthritis classification. The knee X-ray image data sets were obtained from the Osteoarthritis Initiative (OAI) in 2011. The classification was based on the Kellgren-Lawrence (KL) grades, which related to the various stages of OA solidity. The classifier was constructed using manual knee X-rays image classification, indicating the first four KL grades (normal, doubtful, minimal and moderate). Computer-based image analysis was conducted by employing Machine Learning involving various stages — first, preprocessing using Contrast Limited Adaptive Histogram Equalization (CLAHE) and cropping images manually to 400 × 100 dimension; second, feature extraction by using Structural 2 Dimensional Principal Component Analysis (S2DPCA); and the last stage, classifying the images using Support Vector Machine (SVM). The experimental results showed that KL grade 0 could be differentiated from the other grades with accuracy up to 94.33% on Gaussian kernel.