DL-based Rheumatoid Arthritis Prediction using Thermal Images

Abstract

Rheumatoid arthritis, often known as rheumatoid, is an inflammatory condition brought on by the immune system’s malfunction.Various preliminary tests were proposed to predict this chronic illness. This study proposes a deep learning model which can detect the presence of rheumatoid by analyzing the thermal images of a person. For this purpose, the palms of the rheumatoid patients and the control group were scanned to produce a sample of thermal pictures of human hands. The efficiency of this training is then improved by preprocessing the thermal pictures. The CNN-LS TM approach is used to build a deep learning model. Then, to accurately forecast the presence of rheumatoid, this model is trained using thermal pictures. The training’s outcomes are noted and reviewed. Validation comes after training, and the outcomes of the validation are also tabulated. For simpler analysis, the findings are also plotted as graphs. The results show that as the number of epochs rises, accuracy, precision, and recall value all significantly increase. As the number of epochs rises, the loss value also falls. The model is then tested to determine the final values for each parameter after training and validation. The final accuracy score of the model is 92.78, while the loss score is 3.78, which is so minuscule as to occasionally be ignored. The model’s precision is 95.4%, and its recall value is 93.7%. This deep learning model can be utilized as a screening tool for rheumatoidbecause of its improved accuracy and precision values.