Deep Learning Approach to Recognize COVID-19, SARS and Streptococcus Diseases from Chest X-ray Images

Verma, Kamal Kant


Corona Virus Disease (COVID-19) became pandemic for the world in the year 2020. A large numbers of people are infected worldwide due to the rapid widespread infectious virus which is threatening many lives and economic damages. Controlling of this virus becomes challenging for the world due to non-preparedness and less availability of testing kits, necessary medical equipment, and vaccine. Pathological laboratory testing of a large number of suspects becomes challenging. Some existing pathological testing is producing false-negative results. Therefore, this paper aims to develop a method of automatic detection of transmissible diseases through medical image analysis techniques which are based on the radiological changes in the X-ray images. In this paper, a Deep Learning approach is proposed for the fast detection of COVID-19, Streptococcus, and Severe Acute Respiratory Syndrome (SARS) positive cases. In Deep Learning, 2-D Convolution Neural Network (2DCNN) is used to classify graphical features of X-ray image’s dataset of COVID-19 positive, Streptococcus and Severe Acute Respiratory Syndrome (SARS) patients. The proposed approach is implemented on the COVID-chest X-Ray dataset. Experiments produced individual accuracy of COVID-19, Streptococcus, SARS disease and normal person is 100%, 90.9%, 91.3%, and 94.7% respectively. This approach achieved an overall accuracy of 95.73% over four classes. Validation of the proposed approach results has been done using Precision, Recall, and F1-score matrices. From the experimental results, it is proved that the performance of the proposed deep learning approach is quite better as compared to the mentioned state-of-art methods to detect COVID-19, SARS, and Streptococcus disease using X-ray medical imaging.


COVID-19; Pandemic; Convolution Neural Network; Computed Tomography; Medical Image Processing;X-ray; SARS; Deep Learning

Full Text: PDF (downloaded 925 times)


  • There are currently no refbacks.
This abstract viewed 1436 times