A Rapid analysis of the Covid 19 using AI using Chest X-Ray

Abstract

Healthcare systems around the world are overwhelmed by the rapid increase in COVID-19 cases. Because of the limited number of testing kits available, it is not possible for all patients with COVID-19 to be tested using conventional methods (RT-PCR). These tests also have a slow turn-around time and limited sensitivity. Detecting COVID-19 infection via Chest X-Ray can be used to quarantine high-risk individuals while they wait for the test results. X-Ray machines are available in nearly all healthcare facilities. Modern Xray machines are digitalized, so samples don't need to be transported. To prioritize patients for further RTPCR testing, we recommend using chest Xray. This could prove useful in an inpatient setting where the current systems are unable to determine whether the patient should remain in the ward or be isolated in COVID-19. If a patient has a false positive RTPCR, this would allow them to be identified as being at high risk for COVID. Repeat testing would be necessary. In situations where radiologists cannot be reached, we propose using modern AI techniques to detect COVID-19 cases by X-Ray images. This will make it more adaptable. COVID-19AI Detector is the COVID model. This model uses deep neural network to triage patients and determine appropriate testing. On the publicly available covid-chestxray-dataset [2] dataset, our model gives 90.5% accuracy with 100% sensitivity (recall) for the COVID-19 infection. Our results are significantly superior to those of CovidNet [10] which uses the same dataset.