AI Based approaches for identification of COVID and non-COVID Pneumonia

The COVID–19 pandemic affected millions of lives worldwide, disrupting every aspect of economic and social life, presenting unprecedented challenges to the public health sector. Early and accurate detection of the disease is important to contain the pandemic and minimize further loss of lives and livelihood. Artificial Intelligence and Machine Learning (AI/ML), like in many other domains, have also been instrumental in tackling the outbreak of Coronavirus. From detection to treatment, AI/ML has been used by various industries for faster and automated response. RT-PCR being the standard detection technique for COVID-19, takes few hours to provide results. X-ray and CT scan images of the lungs provide a clear representation of the presence of virus, indicating the amount of damage incurred by the lungs. In this paper, an attempt has been made to differentiate COVID-19 CT scan images from that of non COVID-19 using AI/ML. A detailed study has been done on the artificial neural networks used worldwide by various researchers for this classification. Convolution Neural Network (CNN) being the most effective classifier for images, has been used in this study to build a neural network (Covi-net) that can classify COVID-19 and nonCOVID-19 images with 97.6% accuracy.