Predicting Covid-19 Based on Chest X-ray Image Using Attention Guided Contrastive CNN

The COVID-19 pandemic continues to disrupt healthcare systems around the world. Across numerous countries, the 2nd surge is veritably severe. The World Health Organization (WHO) declared this contagion as a global epidemic after it contagion numerous people and claimed in numerous lives around the world. An infection caused by Covid19 illness wreaks annihilation on the mortal pulmonary system, leading to several organ failures and, in the worst-case script, death. Multiple deep learning literacy CNN architecture were used to extract features from chest X-ray, which were also classified as Covid19, Pneumonia, using the chest X-ray as input. Affordable and rapid testing and diagnosis is essential to combat communicable diseases. Currently, testing for Covid19 is expensive and time consuming. A chest x-ray (CXR) may be the fastest, most scalable and non-invasive system. Existing methods are hampered by the limited number of CXR samples available in Covid19. Therefore, inspired by the limitations of open source work in this area, we propose a center of attention contrast CNN for Covid19 detection in CXR images. The proposed system uses contrast loss to learn powerful and important features.Also, the proposed model gives further significance to the infected regions as guided by the attention medium. We compute the sensitivity of the proposed model over the publicly available Covid-19 dataset. It is observed that the proposed AC-Covid-Net exhibits very promising performance as compared to the existing model even with the less training data sets.