Arthritis Detection Using Thermography and Artificial Intelligence

Abstract

An important factor affecting synovial joints is arthritis. Researchers are exploring this issue in search of early arthritis detection methods. The detection of autoimmune disorder has been made possible by several technologies, though early detection is the most important in preventing the immune system from being unable to fight off bacteria and viruses, damaging healthy tissue, leads to painful, swollen, and stiff joints due to inflammation. Thermal imaging, using high resolution cameras of high sensitivity represents a promising early diagnosis method. An innovative method of diagnosing arthritis early is to use Artificial Intelligence (AI), in conjunction with thermal images, enabling excellent predictability levels. This research presents a systematic review of relevant literature using AI and thermography by summarizing their contributions and drawbacks, and identifying research gaps. The literature has studied a variety of artificial neural networks (ANNs) and deep learning models for processing thermo-grams, such as K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Bayes Net, Convolutional Neural Networks (CNN), Convolutional and DeConvolutional Neural Networks (C-DCNN). The number of thermal images varied in previous studies. According to literature study, A number of factors, such as the database, the optimization method, a model and the extracted features, can affect the performance of the neural network. Most studies achieved a classification accuracy range of 70 to 85% as a result of using small samples.