Classification of Long COVID Pulmonary Fibrosis Based on Computed Tomography in Mexican Patients

Abstract

In this study, we focus on addressing the prolonged effects of COVID-19, specifically pulmonary fibrosis, through automated classification of computed tomography (CT) images from patients in León, GTO, Mexico. We employed a convolutional neural network (CNN) VGG16 along with image enhancement filters, such as Meijering and Roberts, to optimize image quality. Prolonged COVID has been linked to complications, including pulmonary fibrosis, underscoring the need for early detection. Our model, supported by a comprehensive dataset, achieved classification accuracy exceeding 97%, successfully distinguishing between patients with and without pulmonary fibrosis. The combination of enhancement filters and CNN VGG16 proved crucial in this success, highlighting the potential of our model for early detection and effective management of pulmonary fibrosis in long COVID patients. This promising approach may have a significant impact on clinical practice, enhancing outcomes and enabling timely interventions. In summary, we present an effective method contributing to timely diagnosis and treatment of pulmonary fibrosis in long COVID patients.