Hand Arthritis Disease Classification using Convolutional Neural Network

Abstract

The prevalence of arthritis is rising worldwide. Psoriatic arthritis, Rheumatoid arthritis and osteoarthritis are the most and frequently affecting arthritis joint chronic diseases. This arthritis disease causes pain, function limitation, and permanent joint damage in the hands and other joints of the body. Plain hand x-ray images are the most commonly used imaging methods for the diagnosis, differential diagnosis, and monitoring of arthritis disease. In this study, the convolutional neural network algorithm was used to obtain hand x-ray images, and classification was made by proposing hand arthritis CNN model. To investigate this research we use 1645 hand arthritis x-ray images in four classes. The data augmentation method was applied during the training of the network. We train 1645 hand x-ray image using 150x150 image size using 50 epoch. After we train, the result of the study were evaluated using deep learning evaluation metrics such as accuracy, sensitivity, specificity, and precision calculated from the confusion matrix of the model. In the classification of psoriatic arthritis, rheumatoid arthritis, osteoarthritis and normal hand x-ray image 91.4% accuracy, 83.2% sensitivity, x94.2% specificity, and 83.2% precision results are achieved respectively. In this study, to develop a computerized method, the CNN algorithm was used to classify the suspected hand x-ray images. In overall we explored the potential of computer aided image processing namely deep learning for hand arthritis detection and screening, and the findings shows that it might be used as a screening tool to classify hand Arthritis in low resource settings.