The sudden increase in COVID-19 patients is a major shock to our global health care systems. With limited availability of test kits, it is not possible for all patients with respiratory infections to be tested using RT-PCR. Testing also takes a long time, with limited sensitivity. The detection of COVID-19 infections on Chest X-Ray can help isolate patients at high risk while awaiting test results. X-Ray machines are already available in many health care systems, and with many modern X-Ray systems already installed on the computer, there is no travel time involved in the samples. In this work we propose the use of chest X-Ray to prioritize the selection of patients for further RT-PCR testing. This can be useful in a hospital setting where current systems have difficulty deciding whether to keep the patient in the ward with other patients or isolate them from COVID-19 areas. It may also be helpful in identifying patients with high risk of COVID with false positive RT-PCR that will require repeated testing. In addition, we recommend the use of modern AI techniques to detect COVID-19 patients who use X-Ray imaging in an automated manner, especially in areas where radiologists are not available, and help make the proposed diagnostic technology easier. Introducing the CovidAID: COVID-19 AI Detector, a model based on a deep neural network of screening patients for proper diagnosis. In a publicly available covid-chest x-ray-dataset [2], our model provides 90.5% accuracy with 100% sensitivity (remember) to COVID-19 infection. We are greatly improving the results of Covid-Net [10] on the same database.
[1]
Ronald M. Summers,et al.
ChestX-ray: Hospital-Scale Chest X-ray Database and Benchmarks on Weakly Supervised Classification and Localization of Common Thorax Diseases
,
2019,
Deep Learning and Convolutional Neural Networks for Medical Imaging and Clinical Informatics.
[2]
Kate Saenko,et al.
RISE: Randomized Input Sampling for Explanation of Black-box Models
,
2018,
BMVC.
[3]
Kilian Q. Weinberger,et al.
Densely Connected Convolutional Networks
,
2016,
2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
[4]
Debdoot Sheet,et al.
Jointly Learning Convolutional Representations to Compress Radiological Images and Classify Thoracic Diseases in the Compressed Domain
,
2018,
ICVGIP.
[5]
Alexander Wong,et al.
COVID-Net: a tailored deep convolutional neural network design for detection of COVID-19 cases from chest X-ray images
,
2020,
Scientific reports.
[6]
Joseph Paul Cohen,et al.
COVID-19 Image Data Collection
,
2020,
ArXiv.