Hybrid Approach for Automated Discrimination Between COVID-19 and Similar Respiratory Diseases

One of the main challenges in accurately diagnosing COVID-19 is its clinical manifestations, which are similar to some respiratory diseases such as viral and bacterial pneumonia, and even influenza during cold seasons. These similarities may lead to misdiagnosis and periodically threaten public health. In order to achieve an accurate and rapid diagnosis of COVID-19 and differentiate it from other respiratory diseases with similar features, this research aims to provide a comprehensive system. This system utilizes artificial intelligence and machine learning methods to analyze lung CT scan images to distinguish COVID19 from other lung diseases. To train this system, data from lung CT scan images of patients from Imam Hussein Hospital in Tehran were used. The data includes three categories of patients: COVID-19 patients, patients with lung pneumonia, and patients with other respiratory conditions. In this study, two main approaches to patient classification were conducted using artificial intelligence and machine learning methods. The results show that deep learning models and hybrid models achieved acceptable performance with accuracies of 98.98% and 99.79%, respectively.