Deep Learning in Radiology:A Transfer-Learning Based Approach for the Identification and Classification of COVID-19 and Pneumonia in Chest X-ray Images

Pneumonia, a prevalent infection of the lungs, often arises from bacterial or viral sources. Its impact is notably significant in underdeveloped and developing nations characterized by high pollution levels, unhygienic living conditions, overpopulation, and limited healthcare infrastructure. Detecting pneumonia early is paramount for ensuring effective medical intervention and improving mortality rates. Radiograph of the chest (CXR) examinations represent the most widely used diagnostic way of pneumonia detection. However, interpreting CXRs is challenging and susceptible to subjectivity. Throughout our study, we have devised an automated system for pneumonia detection from CXR images. To address the data scarcity issue, we employed deep transfer learning techniques on VGG16 and ResNet50 architectures. Our proposed approach achieved impressive accuracy rates of 89.23% and 88.80% for VGG16 and ResNet50, respectively. This performance surpasses that of existing methods, underscoring the effectiveness of our approach.