Desarrollo de una aplicación móvil, para el reconocimiento de covid 19 en imágenes de radiografía de tórax

Abstract

Currently, we are facing a serious health problem due to the rapid spread of the covid-19 virus that does not give truce, at the beginning of the pandemic, it was thought that this virus only attacked elderly people, but with the passing of the months and given the daily infection rates classified by age, it allows us to understand the seriousness of the case and shows the instability and deficiency of the health system of the countries, but mainly in Ecuador where medical resources were not sufficient to meet the high demand for patients infected daily and where, despite efforts and daily sampling, results could not be issued in less time and prescribed to start the recovery process and prevent new infections. It is for this and other reasons that a group of students from the city of Madrid came up with the idea of creating a model that recognizes a coronavirus infection with up to 97% accuracy in X-rays, this research and creation seeks a way to facilitate issuing diagnoses to patients. That is why, motivated by the existence of a neural network model already created and the fundamental role that mobile applications have played in the area of medicine, the idea of processing and training a neural network based on previous research arises. proceeds to train the neural network in the Firebase application, taking the data set of X-ray images from GitHub and developing a mobile application that allows to demonstrate the functioning of the trained network and that this is of great help for medical personnel reducing time of response, diagnostic costs and most importantly efficient medical prescriptions for the treatment of infected patients. The dataset that has been used for the training was obtained from the GitHub repository and is made up of a total of one thousand (1000) images comprising images of chest X-rays of healthy lungs and diseased lungs with pneumonia caused by the Covid-19 virus.