Determining Efficient Machine Learning Techniques for Grading of Knee Osteoarthritis

Abstract

Osteoarthritis (OA) of the Knee is a degenerative joint disease mainly caused due to loss of articular cartilages. The paper introduces an approach to quantify knee osteoarthritis (OA) severity using KL grades. This approach combines EDA (Exploratory Data Analysis), Pre-processing and Feature Engineering techniques. The amount of damage to the knee can be graded using KL scale (0-4). The automated detection of Knee Osteoarthritis (KOA) based on KL grades which corresponds to severity stages has been given in the paper. In the study public dataset from Osteoarthritis Initiative (OAI) has been used to evaluate the proposed approach with very promising results. Different accuracy metrices like F1 score, Receiver operating characteristic curve (ROC), Area Under Curve (AUC) and Precision were used to find the best algorithm amongst the classification models in Machine learning. Random forest and Decision trees algorithms were considered efficient giving an accuracy of 96.9% and 91.6% respectively. Our study is an economically better approach when compared to x-rays for OA detection