Leveraging Machine Learning for Enhanced COVID-19 Detection From Chest X-Ray Images

The emergence of the COVID-19 pandemic, caused by the SARS-CoV-2 virus, underscores the critical need for effective diagnostic methods. One promising approach utilizes deep learning algorithms to analyze lung X-ray images, enabling precise and accurate detection of COVID-19. Implementing this algorithm has the potential to expedite the medical process of recognition and detection, thereby enhancing healthcare systems, particularly in resource-limited regions. This method promises not only rapid and accurate diagnosis but also supports effective viral treatment and efficient management of medical resources. In our study, we employed convolutional neural networks (CNNs) to process chest X-ray images for the identification of COVID-19 cases. The primary stages involved collecting and labeling datasets, followed by feature extraction and recognition using various CNN models, including Xception, VGG16, VGG19, MobileNet, InceptionV3, ResNet50, and DenseNet. Each model was fine-tuned with appropriate hyperparameters to optimize the training process. We evaluated the performance of the models using test data, assessing accuracy, precision, recall, and F1 score to determine the highest classification accuracy. Our experiments revealed that the DenseNet and Xception models delivered superior performance, achieving classification accuracies of 96.88% and 88.89%, respectively. Notably, the DenseNet model exhibited outstanding performance despite being trained with shorter time frames and smaller datasets, showcasing its efficiency and generalization capabilities. These findings demonstrate the potential for accurate and efficient COVID-19 detection, supporting rapid response efforts, enabling large-scale screening, and improving the management of healthcare resources.