Detection of knee osteoarthritis with artificial intelligence-based methods

Osteoarthritis means that the slippery cartilage tissue, which is located on the bones in the joint areas and facilitates the movement of the joint areas, loses its function and wears out. Knee Osteoarthritis means the erosion of the cartilage slippery tissue in the knee joint area. Knee osteoarthritis is a disease whose risk increases with age. In this study, a hybrid model was developed using Histogram of Oriented Gradients (HOG), Darknet53, Local Binary Pattern (LBP), Neighborhood Component Analysis (NCA) feature selection methods. Our dataset used in the experiments includes a total of 1650 knee joint images and consists of 5 classes: Normal, Doubtful, Mild, Moderate and Severe. In the experimental studies, the performance of the proposed method is compared with Convolutional Neural Network (CNN) models. While the accuracy values of Alexnet, VGG16, ResNet50, ResNet50, Xception, NasNetMobile, EfficientNet-b0, Darknet53, MobileNetV2 ESA models are 55.35%, 61.41%, 50.51%, 33.13%, 34.75%, 36.97%, 65.45%, 44.65%, 54.75% respectively, 83.6% accuracy is obtained with the proposed method.