Image-based diagnosis of COVID-19 and Social isolation during Pandemic Situation

Abstract

Over the past nine months, the world is facing a global well-being emergency, with the 2019 COVID-19 virus appearing as a threatening pandemic. In addition to the number of cases and deaths of this pandemic, socioeconomic, political, and psycho-social effects have also been important. While countries have ensured that social elimination is updated to contain the transmission of contamination, billions of people are isolated in their own homes. There is a classification between the afflicted and the doubtful. This social containment results in persistent exhaustion and weariness, which can affect the physical and mental prosperity if it is long enough. In this method, the photographs of a given patient in a voting system are counted as a group. In the two largest datasets of COVID-19 CT analysis with a patient-based break, the method is checked. In a more practical situation in which data comes from various distributions, across dataset analysis is often proposed to test the robustness of the models. The cross-dataset study has shown that in the best assessment case, the generalization power of deep learning models is far from appropriate for the challenge because accuracy decreases from 87.68 percent to 56.16 percent. These findings highlighted that to be seen as a therapeutic choice, the methods aimed at COVID-19 identification in CT images have to be substantially enhanced and broader and more diverse data sets are required in a practical scenario to test the methods.