Convolutional Neural Networks for COVID-19 Diagnosis

The COVID-19 infection has led to uncontrolled destruction to humankind. Health professionals and law and order bodies are the primary people facing challenges to control coronavirus crisis. However, there is a major responsibility on the researchers in investigating the possible Artificial Intelligence based aids that help in diagnosis of coronavirus infection at an early stage, mitigate the spread of coronavirus infection and bring peace and confidence in the people across the world. AI-driven tools show incredible potential to draw high precision interpretation from medical images and extend great support to doctors in decision making. Therefore, AI-powered support systems have been equipped as an integral part of health care center infrastructures. Chest-CT scans and X ray imaging for COVID 19 infection diagnosis aids in evaluating emergent conditions of patient is the necessary as RT-PCR test used for Coronavirus infection detection takes longer time. The examination of CT scan and X -Ray images from suspected patients with infection by an expert radiologist required careful observation and interpretation. With the outbreak in corona virus infection radiologists are subjected to extraordinary workload. Therefore, to reduce workload of radiologists it is necessary to develop an automated tool. In the present COVID 19 crisis there is emergency for development of an automatic tool which is helpful in quick analysis. The research outcomes of the state-of- the-art investigations in this direction indicate that COVID-19 detection models which are developed using the paradigm of deep convolutional neural networks have shown high hopes challenging COVID 19 crisis. In this chapter discusses, fundamentals of convolutional neural networks, transfer learning techniques, provides a list of COVID-19 dataset, COVID- 19 diagnosis systems based on DenseNet, ResNet and SqueezeNet architectures, their comparison, ensemble techniques and research directions. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.