Detection and classification of Covid-19 in CT-lungs screening using machine learning techniques

Abstract

In this study, machine learning is used to detect coronavirus in CT-lung screening. This virus causes serious acute respiratory illness and is very contagious. The WHO has confirmed 19 cases of covid. The case rate is above 12,000,000 and the mortality rate is over 500,000. In the absence of a vaccine, early discovery of this pandemic is the most effective technique to restrict its spread and lessen its consequences, preventing the virus from killing the patient’s lung. While some specialists believe that RT-PCR testing is the best approach for identifying Covid-19 patients, others believe that CT scans of the lungs might be more accurate and less expensive than PCR testing. Many researchers are interested in creating CAD (computer-aided diagnostic) systems to help radiotherapists detect corona virus in the lungs. Thus, the goal of this project is to design a CAD system for identifying and classifying Covid-19 virus in CT images using machine learning. The suggested CAD system comprises four phases: collection of CT Lungs Screening, pre-processing to increase the appearance of the ground glass opacities (GGOs) nodules (initially hazy with feeble contrast), and finalisation. After thresholding and filtering, a collection of typical characteristics is studied to reduce the rate of false negatives in the picture. Second, the affected locations - Regions of Interest (ROIs) - will be detected. This will be done using a modified K-means method. Finally, we will harvest the solid false positives that seem to be GGOs as inputs and targets for the machine learning classifiers, here we will utilise radial basis function (RBF) and support vector machine (SVM).