International Research Journal of Education and Technology

Rheumatoid Arthritis shortly called as RA is a complex systemic autoimmune disorder that affects the internal body tissues and frequently leads to irrepairable structural damage by causing chronic inflammation, mainly in synovial joints. Early detection of rheumatoid Arthritis is   very important so that the problems can be avoided in the future. Outbreaks of rheumatoid Arthritis have increased in white people by 3% - 7% almost every year. Medical assessments, laboratory investigations, and patient self-evaluation are used to determine the disease's activity. Disease progression and responses to treatment also vary substantially, even though there are many alternative treatment options available for different stages of disease. Radiographs of the hands and feet are used to determine the disease's prognosis over the long term which is the input image for our model, also evaluating the X-ray images by trained medical staff requires a lot of time to assist the patient's condition and stages which is a tedious task. In the same way, a timely diagnosis of the illness is crucial for the patient's treatment for a chronic autoimmune illness which is expensive to treat and has a poor survival rate. Machine learning researchers have worked hard to create quick and

precise automatic approaches for diagnosing RA. Deep learning (DL) has given rise to a booming body of academic study as well as industrial applications in the field of medicine. If used properly, deep learning might be very relevant to rheumatology. The ability of deep learning to learn the structure of the underlying data is the key to this effectiveness. Radiographs of the hands and feet are physically examined and graded to determine the extent of joint destruction in Rheumatoid Arthritis (RA).In order to identify and detect rheumatoid arthritis by hand, particularly during its early development or pre- diagnostic stages, an efficient system analysis is required. The goal is to create an intelligent system that can recognize rheumatoid arthritis by utilizing convolutional neural networks (CNN) in deep learning. Our dataset contains images of four different classes of rheumatoid Arthritis such as Synovitis, Pannus, Fibrous Ankylosis, Bony Ankylosis. To remove noise from images, we have applied augmentation (adjustment) techniques such as brightness, zoom, rear, flip, etc. Using augmentation techniques, we are making the dataset clearer and generating new dataset from existing dataset. Based on different values of epochs and other