ABSTRACT
There is enormous attention surrounding COVID-19 these years. Artificial Intelligence are rising in the medical field. However, the next revolutionary steps lie in the upsurge of deep learning methodology. In this study, we propose a optimal Resnet50 based deep learning network for proper CT diagnosis of the COVID-19. The learning model was then adopted on the CT scan images with COVID-19 dataset (provided by Zhuhai HuiYu Medical Technology) and eventually achieves the autonomous diagnosis of new images by reading and analyzing more than 5000 CT pictures. The preliminary results of the core system show that the detecting sensitivity of the algorithm is 98.7, and the accuracy of the COVID-19 case detection is up to 97.4%. The CT detection for one COVID-19 case is expected to take 5.3 seconds, which greatly reduces the working intensity of doctors and the number of misdiagnoses significantly. In addition, the system can be extended to other more extensive medical image perception in the future. © 2022 IEEE.