Detection of COVID Using Deep Learning: A Review

ABSTRACT

As we all are aware of the fact that India's population is increasing expeditiously, automatic diagnosis of diseases is now crucial topic in medical sciences. Coronavirus has expanded massively, and it is among the one of the most frightful and dangerous infection in latest years. The deadly virus was found in China first, and then, it mutated throughout the world. Hence, automated illness identification provides results that are uniform and quick, and thus, mortality rate can be reduced. Most of countries including ours (India) suffers from lack of testing kits whenever new wave of COVID hits. Therefore, many researchers worked on various deep learning based, machine learning-based approaches for diagnosis of this virus using X-rays and CT scans of lungs. So far, it has affected over 50.9 crore people and caused the deaths of 62.2 lakhs people. Here, in this study, comprehensive survey of fifteen studies is presented where various deep learning, and transfer learning approaches are compared for their efficiency and accuracy. The goal of study here is to inspect and analyse various deep learning models including transfer learning models used, also explore the datasets used, preprocessing techniques used, and compare these models to find which model provide us with optimal and best results. The study can help in smooth implementation of the suggested work in future which can be further, then fine-tuned to get the best results possible. Deep learning provides an easy solution to the COVID problem as they perform best in detection and evaluation. It is found during this study that CNN model hybridized with other models provide better accuracy then CNN alone. Ensemble learning methods also improves the accuracy. Also, before training any model dataset acquired need to be preprocessed. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.