Classification of Covid-19 Chest X-Ray Patients using Artificial Neural Network

In recent years, there has been an increase in the COVID-19 outbreak, which appears to be worsening significantly due to the shortage of rapid testing kits. As a result, it is critical to develop automated systems for COVID-19 detection based on radiological images in order to detect the presence of disease. A dry cough, sore throat, and fever are the most common signs and symptoms of COVID-19. According to Covidnow statistics, the local number of cases is 4,810,082, the number of recovered cases is 4,788,889, and the number of deaths is 36,387. The swap test for the Polymerase Chain Reaction Test (PCR) takes a long time because the test must be sent to the lab to obtain the result. Early detection is expected to contribute to a reduction in the rate of viral transmission. The Artificial Neural Network (ANN) methodology was discovered to be one of the most basic methods for dealing with complex situations. ANN considers classification one of the most dynamic areas of research and application. The method for classifying COVID-19 using ANN was proposed in this study. Models of complex natural systems with many inputs may be easier to use and more accurate whenever ANNs are used. There are 32 main features extracted from segmented lung X-ray images and used as neural network inputs for the classification process, which include shape, texture, colour, and moment. The MLP network was presented in order to classify the state of COVID-19 (COVID-19 or Normal Chest X-ray) using Levenberg-Marquardt, Bayesian Regularisation, and Scaled Conjugate Gradient. Overall, MLP-LM achieves the highest testing accuracy of 99.93% for 11 hidden nodes when using all 32 input features. As a result, the MLP network proves suitable for detecting COVID-19 from lung X-ray images.