Leveraging X-Ray and CT Scans for COVID-19 Infection Investigation Using Deep Learning Models: Challenges and Research Directions

Abstract

The present health situation in the world has proven that COVID-19 pneumonia is life threatening lung infection and is contagious. Health professionals are facing a big challenge in facilitating the treatment needs and rigorously carrying out the research to invent a vaccine. There is a difficulty in COVID-19 diagnosis through RT-PCR pathological test. The test needs few days for the results, and therefore, immediate treatment plan is not possible which may further deteriorate patient’s health. Deep learning techniques are used in the medical domain which have outperformed expert opinions in some cases. There is a significant amount of research going on to develop automatic tools for the analysis of chest X-ray and chest CT scans from COVID-19 suspected patients. Convolutional neural networks are the most powerful models which are used for X-ray image analysis for detection of COVID-19 pneumonia lung infection. In this paper, state-of- the-art survey on the analysis of chest X-ray and CT-scan images for COVID-19 pneumonia detection using deep learning methods is presented. The contributions of this work include identification of pre-trained deep learning models that have yielded best results, strategies to counter imbalanced COVID-19 dataset and identification of authorized COVID-19 positive datasets that are used in research work.