NEURAL NETWORK AIDED OPTIMIZED AUTO ENCODER AND DECODER FOR DETECTION OF COVID-19 AND PNEUMONIA USING CT-SCAN

ABSTRACT

Most of the countries in the world are now fighting against Covid-19 and many of the people are losing their life because of the less immunity or due to the late diagnostics and it is especially in the case of old age people and people with other medical issues. The concept of early detection of disease is really important in the case of the Covid-19 scenario because along with the infected people, the other people who are in close contact with the infected persons will also have life risk. During this pandemic, pneumonia and Covid-19 people suffers from almost the same symptoms. So, the proposed work designs an automated system that can perform multi-classification on general health, pneumonia and Covid-19 through Chest X-Rays by designing an optimized auto encoder- decoder network. Most of the earlier approaches which are used to perform the binary classification couldn't differentiate the Covid-19 and Pneumonia effectively because the traditional CNN extract the high level features, which are similar in case of COVID-19 & Pneumonia. These two have variations in the case of low level features. The major focus of this paper is to construct a hyper-parameterized auto encoder-decoder system that can help the user to detect level of lung infection. The level of infection helps the model to accurately classify the model. This method helps doctors and other medical-related people with the early diagnosis of disease. © 2022 Little Lion Scientific.