Severity Stage of Osteoarthritis Prediction in Knee Joint using Transfer Learning Approach

Knee Osteoarthritis (KOA) is a progressive disease of the joints that damages cartilage, bones, and surrounding tissues. Osteoarthritis can cause pain, stiffness, and decreased range of motion in the affected joints. Early prediction of arthritis also reduces the chance of complications and helps prevent further damage to the body's afflicted joints. This study has proposed a transfer learning model, DenseNet201, that predicts the severity of osteoarthritis stage using knee X-ray images, which include left and right-span X-ray images of several joints to train the model. Additionally, this work compares five classes versus three classes of the severity stage of osteoarthritis in knee joints and analyzes the performance of the model. The Adamax optimizer is used to optimize and achieve higher accuracy results. The optimizers have different values for recall, accuracy, precision, and F1-score. After the comparison of five classes, this work concludes three classes exhibit improved results for the prediction of osteoarthritis that achieve higher accuracy, such as 98.86% and 96.24% with five classes, respectively.