Medical Image Retrieval using Stacked Autoencoders : COVID-19 Application

Abstract

COVID-19 is a recently discovered infectious disease caused by the coronavirus known to cause respiratory infections in humans. This pandemic is spreading rapidly around the world, causing multiple damages in different areas. In this graduation project, we are interested in the recognition of this disease using med- ical images. For this purpose, we present an application dedicated to epidemiol- ogists for the investigation of the Patient 0 infected and establish the propagation path in different areas of the country. A Content Based Medical Image Retrieval (CBMIR) system based on stacked-encoder networks is proposed, our model is dedicated to search for target COVID Chest X-Ray images using similarity mea- surements learned through an image database of different pathologies as SARS and other viral or bacterial species of pneumonia diseases.