Artificial Intelligence in Lung Cancer Diagnosis: A Comprehensive Review of Ml and Dl Approaches

Abstract

Lung cancer remains one of the leading causes of cancer-related mortality worldwide, necessitating accurate and timely diagnostic approaches. This comprehensive literature review examines the application of deep learning and machine learning techniques for lung cancer detection during the period 2019-2023. Based on a systematic analysis of 282 unique papers filtered to 30 highly relevant studies, this review synthesizes current methodologies, datasets, and performance metrics in the field. The analysis reveals that convolutional neural networks (CNNs) and their variants, particularly 3D architectures and hybrid models, have emerged as the dominant approaches, achieving accuracies ranging from 71% to 99% across various datasets. The LIDC-IDRI dataset has become the de facto standard for benchmarking, while imaging modalities primarily focus on CT scans. Key findings indicate that ensemble methods, transfer learning, and attention mechanisms significantly enhance detection accuracy and reduce false positive rates. However, challenges remain in model interpretability, dataset diversity, and clinical deployment. This review provides a structured analysis of methods, datasets, and performance metrics, offering insights for researchers and practitioners advancing automated lung cancer detection systems.