Lung-Related Diseases Classification Using Deep Convolutional Neural Network

Abstract

Accurate diagnosis is a crucial first step in the management and treatment of lung
\ndiseases, which include infectious diseases such as COVID-19, viral pneumonia, lung
\nopacity, tuberculosis, and bacterial pneumonia. Despite these conditions sharing similar
\nmanifestations in chest X-ray images, it is imperative to correctly identify the disease
\npresent. This study, therefore, sought to develop a convolutional neural network (CNN)-
\nbased model for the classification of lung diseases. Four distinct CNN models, namely
\nMobileNetV2, ResNet-50, ResNet-101, and AlexNet, were rigorously evaluated for
\ntheir ability to classify lung diseases from chest X-ray images. These models were tested
\nagainst three classification schemes to examine the impact of high interclass similarity:
\na 4-subclass classification (COVID-19, viral pneumonia, lung opacity, and normal), a
\n5-subclass classification (COVID-19, viral pneumonia, lung opacity, tuberculosis, and
\nnormal), and a 6-subclass classification (COVID-19, lung opacity, viral pneumonia,
\ntuberculosis, bacterial pneumonia, and normal). The retrained ResNet-50 architecture
\nyielded the best results, achieving a classification accuracy of 97.22%, 92.14%, and
\n96.08% for the 6-subclass, 5-subclass, and 4-subclass classifications respectively.
\nConversely, ResNet-101 demonstrated the lowest classification accuracy for the 6-
\nsubclass and 5-subclass classifications, with 78.12% and 79.49% respectively, while
\nMobileNetV2 had the lowest accuracy for the 4-subclass classification, with 88.89%.
\nThese results suggest that, despite high interclass similarity, the ResNet-50 model can
\neffectively classify lung-related diseases from chest X-ray images. This finding
\nsupports the use of computer-aided detection (CAD) systems as decision-support tools
\nin the early classification of lung-related diseases.