Automated Recognition Of COVID-19 Cases From Chest X-Ray Images Using Multi-Image Augmented Deep Learning Model

Document Type : Primary Research paper

Authors

1 Bannari Amman Institute of Technology,

2 Bannari Amman Institute of Technology

Abstract

Looking at the global pandemic that had a huge impact on human lives since the
last several months where people are losing their lives, people are losing their jobs and it is
having a devastating effect on human life. There are frontline health Care community
workers for identifying various solutions to minimize this impact. On the other hand there
is a different set of people in the data science community who are looking at various different
technology related solutions to assist the frontline healthcare community with their
solutions. The recognition of covid-19 positive belongings at the early stage in order to
preclude the blowout is most crucial. The RT-PCR is a technique to analyse the occurrence
of covid-19 by taking and nasal or throat swab from the patients which perceives the
capacity of antibodies which are formed by the resistant system. This is the unintended
technique of analysis the presence of virus and the antibodies can display amid 7 to 28
days afterward the contagion. But the radiologists proved that the presence of covid-19
virus can be detected by using the changes that has been occurred in chest X-ray images.
Due to the limited number of radiologist present across the world there is a challenge for
determination of covid-19 using x-ray images. This work aims to represent a framework in
order to automatically diagnose covid-19 in x-ray images using multi image augmented
deep learning model. The filtered images from the CNN produces discontinuity in their
information which can be resolved using multi image augmentation technique thereby accumulative
the amount of images for preparation of the CNN model. This simulation has
been done through to the databases which are available publicly and the proposed technique
provides higher accuracy, sensitivity and specificity.

Keywords