Deep Learning Based Thoracic Disease Detection and Localization in Chest Radiographs

In the area of medical imaging analysis, deep learning has demonstrated significant promise, notably in the detection and treatment of thoracic disorders. With a focus on COVID-19, lung opacity, and viral pneumonia, this research study attempts to examine the working and use of deep learning approaches for the classification, segmentation, and localization of thoracic illnesses. In order to accurately classify, segment, and localize diseases, the study uses appropriate dataset of posteoranterior chest radiographs and provides a deep learning-based architecture that blends convolutional neural networks (CNNs) with U-Net. The findings show how deep learning may be used to help medical practitioners identify and analyze thoracic disorders. The results of classification and segmentation demonstrate an average validation accuracy of 98.5%.