Rheumatoid Arthritis Detection method based on Mask R-CNN

Rheumatoid arthritis is a chronic autoimmune rheumatic disease with the feature of joint erosion. The disease mainly occurs in middle age, and women are more susceptible than men. In clinical, radiologists mainly use hand radiographs to observe the appearance of joints to determine the degree of disease. In order to reduce the pressure of long-term repetitive work of radiologists and improve screening efficiency, we propose a detection method based on Mask RCNN. We use an in-house dataset that contains 99 images and is annotated by professional rheumatoid arthritis doctors. Using this method, we can quickly identify the diseased joints from all hand joints and determine their exact positions. This network can automatically detect diseased joints with higher detection accuracy and faster detection speed than other methods.