Deep learning in Covid-19 detection and diagnosis using CXR images: challenges and perspectives
S. Suganyadevi, V. Seethalakshmi, K. Balasamy, N. Vidhya
Abstract
The COVID-19 pandemic is endangering the health and lives of people in 223 nations and territories. When dealing with the Coronavirus, early discovery and isolation are the most important measures to take. X-rays, computed tomography and magnetic resonance imaging can show the presence of Covid-19, making infection detection a cinch. Chest X-rays (CXR) of people infected with Covid-19 are shown to have certain abnormalities and it is one of the frequently used imaging modalities. The digital twin monitors people's data, compares them to documented patterns, and analyses illness symptoms. Furthermore, the data might be used to create a digital modelling of a typical healthy patient, which aids in the definition of new healthy criteria. There are a variety of ways to use deep learning to find diseases in X-rays from previous efforts. Initially, 6,000 chest X-rays were collected from publicly available sources. These pictures will be recognized by a radiologist as evidence of Covid-19 sickness. Approximately 15% of children worldwide are killed by pneumococcal disease. Recognized and sorted pneumonia, healthy, and Covid-19 all come from the modified VGG16 deep learning (DL) technique. The convolutional neural network (CNN) is a deep neural network that incorporates both external and internal characteristics and is used by the identification model to identify pixels. The findings indicate that medical professionals should reconsider the use of X-ray images in the treatment of certain diseases and that more research into evaluating X-ray technology is required. The suggested detection approach performed better when compared to chest X-rays for pneumonia and non-pneumonia conditions.