Identificação do sars - cov - 2 e pneumonia viral a partir de imagens de raio - X de pulmões com tensorflow, Keras e Deep Learning

Abstract

The pandemic caused by the SARS-CoV-2 virus has resulted in more than 555 million confirmed cases and more than 6.35 million deaths worldwide, triggering a massive epidemic. Pneumonia is also an example of diseases considered to be the main causes of death in many countries, such as viral pneumonia, which is a type of infection that affects the lungs and is caused by viruses, usually the same ones that cause colds and flu, or even SARS-CoV-2. In the first part of this work, it was discussed how SARS-CoV-2 and Viral Pneumonia can be detected in chest X-rays of patients, how the datasets were reviewed and then how to train a deep learning model using o TensorFlow and Keras to predict SARS-CoV-2 and Viral Pneumonia in patient x-ray image dataset. The objective of this work was to develop and train a personalized deep learning model, evaluating convolutional neural network architectures, to determine which of the classes the patient's lung falls into: covid, normal or viral pneumonia in a dataset of images of x-ray, using the Python programming language with Tensorflow, Keras and Deep Learning. The training model based on the VGG19 architecture achieved better performance in image identification and classification. It is concluded that a training model with tensorflow and keras using deep learning may be able to learn and identify patterns