Lung Disease Detection and Classification Based on AI, Deep Learning and Machine Learning: A Comprehensive Survey

Abstract

Background: Lung diseases significantly contribute to substantial global morbidity and mortality, which pose considerable diagnostic challenges. Manual analysis of medical images like chest X-rays and CT scans is often subject to human error, requires specialized expertise, and is time-consuming. To build fully automated systems that analyze the information contained in medical images, robust and efficient algorithms are required. In response, artificial intelligence (AI), especially deep learning (DL), has become an innovative approach that shows excellent results for the automation of lung disease detection and classification. Objective: Following a strict systematic literature review methodology, this study aims to provide an extensive survey of artificial intelligence, machine learning, and deep learning approaches for the detection and classification of lung diseases. Methods: The survey discusses over 30 reported approaches in the field. These approaches have been categorized and evaluated based on the methodology and reported results. We also discuss relevant datasets and benchmarking that have been used in state-of-the-art models and have undergone peer review. Results: The survey reveals a strong trend towards powerful ensemble and hybrid architectures like ResNet, VGG, RVCNet, Vision Transformers (ViTs), etc., and their variations. These models have achieved high accuracy for detecting and classifying various diseases. Conclusions: This work highlights the advancement and recent progress in using AI and deep learning to detect and classify lung diseases while addressing identified gaps and the limitations related to dataset availability, standardization, and imbalance, in addition to model transparency and the computational cost, providing a prospective roadmap for further research or implementation.