Pulmonary and cardiac diseases represent one of the greatest challenges to public health, accounting for a significant global mortality rate, a scenario that has been further aggravated by the COVID-19 pandemic, which has highlighted the importance of early and accurate diagnoses. In this context, chest radiography stands out as one of the most effective methods for detecting these pathologies, as it allows a detailed analysis of the rib cage, lungs, and heart, providing crucial information for diagnosis and clinical follow-up. This work proposes a comparative analysis between four object detection models — YOLOv5, YOLOv8, Faster R-CNN, and RetinaNet — with the aim of evaluating which one presents the best performance in accuracy and sensitivity in identifying lung and heart diseases in chest X-ray
images. The research examines the specific characteristics of each model, considering its
effectiveness in identifying various pathologies, such as atelectasis, cardiomegaly, effusion, infiltration and pneumonia, and explores evaluation metrics, such as accuracy, sensitivity and false positive rate, to determine which model stands out in clinical practice. The expected results aim to contribute to the advancement of automated detection of these diseases, offering a solid basis for the implementation of artificial intelligence technologies in clinical settings, with the aim of improving diagnostic accuracy and, consequently, patient outcomes.
|
