Automatic Detection of Knee Joints and Quantification of Knee Osteoarthritis Severity using Modified Fully connected Convolutional Neural Networks

Knee Osteoarthritis (OA) is an extremely common and degenerative musculoskeletal disease worldwide which creates a significant burden on patients with reduced quality of life and also on society because of its financial impact. Therefore, technical try and efforts to reduce the burden of the disease could help both patients and society. In this paper, an automated novel method is proposed with a supported combination of joint shape and modified Fully connected neural network (FCNN) based bone texture features, to differentiate between the knee radiographs with and without osteoarthritis. Moreover, an endeavor is formed to explain the bone texture using CNN. Knee radiographs from Osteoarthritis Initiative (OAI) and Multicenter Osteoarthritis (MOST) datasets are utilized in this paper. The proposed models were trained on 8000 knee radiographs from OAI and evaluated on 3500 knee radiographs from MOST. The results demonstrate that fusing the proposed shape and texture parameters achieves the state-of-the art performance in radiographic OA detection yielding area under the ROC curve (AUC) of 98.75% accuracy.