Uma proposta de adaptação do Software Fibernet com foco no reconhecimento de padrões para pacientes com Covid-19

Abstract

The COVID-19 pandemic highlighted the need for faster and more efficient methods to support the diagnosis of respiratory diseases. In this context, Artificial Intelligence, particularly through deep learning techniques, has shown promising results in the automated analysis of medical images. This study aims to propose an adaptation of the FiberNet convolutional neural network for the classification of chest X-ray images, focusing on the identification of COVID-19 cases. The COVID-QU-Ex dataset was used, containing chest radiography images from COVID-19 patients and healthy individuals. The original FiberNet architecture was analyzed and modified, including adjustments in convolutional layers and hyperparameters to improve feature extraction from the images. Model training was performed using the 10-fold cross-validation technique, with the dataset divided into training, testing, and validation subsets. Model performance was evaluated using accuracy, precision,sensitivity, specificity, and F1-score metrics derived from confusion matrices. The experimental results showed that the adapted model achieved an average accuracy of 95.5%, indicating a high capability in classifying chest radiographic images. The findings suggest that the adapted FiberNet model presents satisfactory performance in detecting COVID-19 from chest X-ray images and may serve as a supportive tool for medical diagnosis.