A Systematic Literature Review of Convolutional Neural Network Architecture for Lung Disease Detection

Lung disease is a severe threat to the respiratory system and can be fatal if not treated immediately. Common examples include Pneumonia, Covid-19, Lung Cancer, and Tuberculosis. Early detection of lung disease is essential to prevent death. Medical imaging techniques, such as X-Ray, Computed Tomography (CT), and Magnetic Resonance Imaging (MRI), play an essential role in diagnosing disease by providing visual information to radiologists and doctors. The use of the CNN architecture for lung disease detection has been increasing recently, with various types of CNN architectures being published separately. Thus, not all types of CNN architectures can be exposed. This literature review aims to identify and analyze the latest research on using CNN architecture in medical image analysis for lung disease detection. Forty-seven research articles posted from January 2019 to May 2023 were retained and selected for further exploration based on predefined inclusion and exclusion criteria. A systematic literature review is a method for identifying, evaluating, and interpreting all available study material to answer a research question. Analysis of the reviewed papers highlights the popularity of the CNN architecture for early detection of lung disease. Among the nineteen identified CNN architectural models, VGG-16 and ResNet50 emerged as the most commonly used for lung disease detection. Pneumonia and COVID-19 are among the lung diseases being studied due to their high global incidence rates, requiring early detection to expedite diagnosis and clinical decision-making. Chest X-ray emerged as the most widely used imaging modality for lung disease detection using CNN due to its accessibility, affordability, and non-invasiveness. While the reviewed studies show high accuracy, optimal performance may vary depending on the dataset, CNN architecture used, and other parameters.