Abstract
In this work we implement a COVID-19 infection detection system based on chest X-ray images with uncertainty estimation. Uncertainty estimation is vital for safe usage of computer aided diagnosis tools in medical applications. Model estimations with high uncertainty should be carefully analyzed by a trained radiologist. We aim to improve uncertainty estimations using unlabelled data through the MixMatch semi-supervised framework. We test popular uncertainty estimation approaches, comprising Softmax scores, Monte-Carlo dropout and deterministic uncertainty quantification. To compare the reliability of the uncertainty estimates, we propose the usage of the Jensen-Shannon distance between the uncertainty distributions of correct and incorrect estimations. This metric is statistically relevant, unlike most previously used metrics, which often ignore the distribution of the uncertainty estimations. Our test results show a significant improvement in uncertainty estimates when using unlabelled data. The best results are obtained with the use of the Monte Carlo dropout method.
Highlights
The COVID-19 pandemic is putting significant pressure on governmental health systems, as the number of cases grows exponentially [1]
In this work we focus in the measurement and improvement of uncertainty estimations for a deep learning model designed to identify COVID-19 infection using X-ray images
We explore the impact of semi-supervised deep learning in the reliability of the uncertainty estimations for COVID-19 detection, using a common deep learning architecture
Summary
The COVID-19 pandemic is putting significant pressure on governmental health systems, as the number of cases grows exponentially [1]. The availability of medical staff is lowered as they get infected by the virus, reducing the overall capacity of hospitals and clinics [1]. The usage of medical imaging can be an alternative tool when other. The usage of computed tomography and X-ray based tests for COVID-19 detection has been studied in [4]–[6], reporting mixed sensitivity and accuracy in the case of X-ray imaging based solutions. The usage of X-ray images is ubiquitous, as this technology is usually cheaper and more widely available [7]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IEEE access : practical innovations, open solutions
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
