Deep Learning-Based Health Care System Using Chest X-Ray Scans for Image Classification

The COVID-19 pandemic has significantly impacted the healthcare systems, other societal systems, and the global economy. The COVID19 virus of the twenty-first century has claimed millions of lives globally in less than two years. Pneumonia is a potentially fatal bacterial disease that affects one or both lungs in humans and is frequently caused by the bacteria Streptococcus pneumonia. Chest radiographic imaging is a precise diagnosis that can be made since the infection affects the patient’s lungs. In this study, the dataset consists of three groups such as COVID-19, normal, pneumonia and viral pneumonia. Deep learning approaches for image classification identify image data, generate results, and classify images for disease identification. Deep neural networks perform the most important aspect in medical image recognition after turning the raw image into a format that can be interpreted by a model, hence pre-processing of the raw image is required. This study’s models were formed from pre-trained CNN models such as VGG, ResNetV2, Dense Net, Xception, Mobile Net, MobileNetV2 and MobileNetV3 versions. The suggested model uses the performance validation of different models, which are summarized in the form of accuracy, precision, recall, F1-score, and AUC. This enables quick diagnosis and aids in differentiating COVID-19 from several types of pneumonia. The MobileNetV3Small model achieved the highest classification accuracy for COVID-19 at 98.57%, and the MobileNetV3Large model achieved the highest classification accuracy for normal, at 99.09%. The DenseNet201 model achieved the highest model classification accuracy for pneumonia (viral pneumonia) at 97.14%.