Abstract
Osteoarthritis (OA) is considered one of the most challenging arthritic disorders due to its high disease burden and lack of effective treatment options that can change the course of the disease. Knee osteoarthritis (KOA) reduces people's quality of life and shortens their daily activities. Therefore, early detection of KOA dramatically impacts patients' quality of life. This study developed an artificial intelligence-supported system to detect KOA. In the developed system, firstly, the images in the original dataset were denoised with a Gaussian filter. Then, feature maps were extracted from both the original and Gaussian applied datasets with the DenseNet201 selected from eight different pre-trained models, and these two feature maps were concatenated. In this way, it is aimed to bring together different features of the same image. Then, feature selection was made using the neighborhood component analysis (NCA) method for the developed system to produce more successful results, and the optimized feature map was classified into six different classifiers. As a result, a high accuracy rate of 85% was achieved in the proposed model. This value is promising for the automatic diagnosis of KOA with computer-aided systems. As a result, a high accuracy rate of 85% was achieved in the developed system of the support vector machine (SVM) classifier. The proposed model was more successful than the other models used in the study.