Classification of lung cancer computed tomography scans using deep networks: A review

Abstract

Lung cancer continues to be the leading cause of cancer-related deaths worldwide, with its mortality rate steadily increasing. Early detection is crucial for improving survival rates, yet the overwhelming workload on radiologists and the shortage of specialists make accurate and timely diagnosis challenging. The large volume of medical images from CT scans, MRIs, and X-rays further complicates the diagnostic process, increasing the likelihood of errors or delays. To address this issue, researchers have focused on developing automated systems that assist in lung cancer detection and classification. This study explores various techniques, including computer-aided detection (CADe) and computer-aided diagnosis (CADx) systems, which utilize medical imaging to identify lung nodules and classify them as benign or malignant. A key objective of this research is evaluating different classifiers to determine the most effective model for accurate classification. Among the models studied, Convolutional Neural Networks (CNNs) have shown the best performance in distinguishing malignant from benign tumors due to their ability to extract complex patterns from medical images. Advanced CNN architectures such as ResNet, VGGNet, and EfficientNet outperform traditional classifiers in terms of accuracy and efficiency. The study also examines segmentation techniques, feature extraction methods, and classification challenges, proposing hybrid AI models and improved data augmentation strategies to enhance diagnostic precision. By addressing these critical aspects, this research aims to develop a robust and automated lung cancer diagnostic framework that enhances early detection, supports radiologists, and improves patient outcomes.