Machine learning model based on Gary-level co-occurrence matrix for chest Sarcoidosis diagnosis

Sarcoidosis is often misdiagnosed and mistreated due to the limitations of radiological presentations. With the recent emergence of COVID-19, doctors face challenges distinguishing between the symptoms of these two diseases. As a result, people are adapting to new practices such as working from home, wearing masks, and using disinfectants. The similarity in symptoms between sarcoidosis and COVID-19 has made it difficult to differentiate between the two conditions, potentially impacting patient outcomes. The diagnostic process for distinguishing between them is time-consuming, labor-intensive, and costly. Researchers and medical practitioners have gained significant attention to computer-aided detection (CAD) systems for sarcoidosis using radiological images to address this issue. This study uses machine learning classifiers, ensembles, and features such as Gray-Level Co-occurrence Matrix (GLCM) and histogram analysis to identify lung sarcoidosis infection from chest X-ray images. The proposed method extracts statistical texture features from X-ray images by calculating a GLCM for each image using various stride combinations. These GLCM features are then used to train the machine learning classifiers and ensembles. The research focuses on multi-class classification, categorizing X-ray images into three classes: sarcoidosis-affected, COVID-19-affected, and regular lungs, as well as binary classification, distinguishing sarcoid-affected cases from others. The proposed method, known for its simplicity and computational efficiency, demonstrates significant accuracy in identifying sarcoidosis and COVID-19 from chest X-ray images.