Classification of the stage of knee osteoarthritis using image processing and deep learning algorithm on direct X-Ray of the knee joint

Every day, new ones are added to the benefits of technology, which has developed and continues to develop until today, in facilitating and accelerating human life. It has become imperative to avoid close contact in human relations and communication, especially after the Covid-19 during pandemic experienced worldwide in 2020. During this pandemic, remote work has been carried out in various public or private institutions for as long as the opportunities and service types allow. The technology used in all areas of our lives is also widely used in diagnosis and treatment applications in order to solve many complex problems that require high accuracy and precision in various applications in the field of medicine. Although the close contact between the health worker and the patient in the health sector is never zero, it is possible to minimize it. In the thesis study conducted for this purpose, a program was developed that can help physicians to make faster and more accurate decisions in the diagnosis phase of knee prosthesis surgeries performed in the field of orthopedics in the health sector by using the classification algorithm. In the study, the classification of the knee osteoarthritis stage in the knee joint is performed by using the deep learning method from the knee joint direct radiographs (X-Ray) and the convolutional neural network (CNN). The widely used Kellgren-Lawrence (KL) scale is used to classify the stage of knee osteoarthritis. In the study, 1464 X-Ray single images obtained from the right and left knees of 732 patients and healthy individuals aged between 45 and 75 are used. While the overall success rate in the prediction values is 99%, the precision values are; 98% accuracy for stage 0 and stage 1, and 100% accuracy for stage 2, stage 3 and stage 4.