TOWARDS ACCURATE COVID-19 VARIANT DETECTION: A DEEP LEARNING APPROACH ON CHEST CT DATA

Abstract

The global COVID-19 pandemic, which started in late 2019, and its subsequent waves have put an unparalleled burden on healthcare systems worldwide, necessitating the rapid development of diagnostic tools and treatments. The prompt identification and accurate classification of COVID-19 patients is critical to stopping the virus's spread and improving patient outcomes. This paper takes up the challenge and embarks on a groundbreaking journey to provide an innovative and practical answer utilizing deep learning, a subset of artificial intelligence.The proposed study suggests a classification method to classify the COVID-19 and other cases from the chest image dataset. There are three steps of proposed methodology, in first step a dataset of CT scan chest images is downloaded from the Kaggle website. The second step of our proposed model is feature extraction and image classification using different deep learning models MobileNetV2, ResNetV2 and proposed a Custom CNN model for better performance. In third step, proposed methodology is detecting Covid-19, Pneumonia and Normal cases from the given image’s dataset. Finally, a comparative analysis is done between the proposed and other well-known techniques on the basis of training and testing accuracy as well as loss. The comparative results show that our proposed model performed well with testing accuracy of 96% percent and give minimum loss