Early Screening Of Covid-19 Pneumonia Patients Using X-Rays By The Developed Artificial Intelligence System

Abstract

The global pandemic of Covid-19 has affected millions of lives . Any technical tool that allows for rapid screening of COVID-19 infection will help Healthcare practitioners to benefit greatly from precision of the results. The most commonly used clinical technique for the diagnosis of COVID-19 is the reverse transcription polymerase chain reaction (RT-PCR), which is expensive, less sensitive, and involves specialised medical staff. X- ray imaging is a simple tool and it’s possible to use it as a good alternative for COVID-19 diagnosis. The research aims to prove that Artificial Intelligence (AI) could help detect COVID-19 from chest X-ray (CXR) images more quickly and accurately. The training and testing datasets were obtained from National Institute of Health(NIH) through kaggle.com. A Transfer learning technique was used with the help of image augmentation to train and verify many pretrained deep Convolutional Neural Network(CNN). The system is trained to classify the data into two different modules: i)Pneumonia affected and Normal ;(ii) Covid-19 Positive and Covid- 19 Negative with image augmentation. Furthermore the patients with non-pneumonia lungs who tested negative for Covid-19 will be given a second diagnosis based on their symptoms. Using Natural Language Processing(NLP) the probability of the patient suffering from Covid-19 can be predicted. For validation dataset random CXR images from the Veteran's Administration (VA) Picture archiving and communication (PAC) system were obtained .Our system's high accuracy can assist in increasing the speed and detection of COVID-19 diagnosis. This will be beneficial in this pandemic, where the risk of illness and preventive measures are at odds with available resources.