AI-Based Diagnosis of Small Pulmonary Nodules on CT Imaging: Current Status and Future Prospects

Abstract

The early and accurate diagnosis of small pulmonary nodules (typically defined as nodules with a diameter ≤10 mm) is crucial for improving the survival rate of lung cancer patients. However, their detection and differentiation pose numerous clinical challenges. Artificial intelligence (AI), particularly deep learning technology, offers promising solutions to these problems. This paper focuses on recent significant international and domestic research achievements supported by clinical validation data, systematically reviewing and analyzing the research progress, core technical methods, clinical application status, and key challenges of AI-based diagnosis of small pulmonary nodules using computed tomography (CT) images over the past decade. It also explores future development trends. Research findings indicate that AI has made remarkable progress in the detection, segmentation, and differentiation of benign from malignant small pulmonary nodules. Deep learning models such as convolutional neural networks (CNNs) and Transformers have continuously evolved, significantly enhancing diagnostic efficiency and accuracy, and to some extent assisting clinical decision-making. Nevertheless, AI still faces substantial challenges in the diagnosis of small pulmonary nodules. At the technical level, these include difficulties in accurately identifying tiny lesions, scarcity of high-quality annotated data, and limitations in model interpretability and robustness. At the application level, challenges involve effectively integrating AI into clinical workflows, gaining physician acceptance, addressing regulatory and ethical lag, and considering cost-effectiveness. Future directions for AI-empowered small pulmonary nodule diagnosis include developing data-efficient and trustworthy AI algorithms, advancing multimodal information fusion, building intelligent diagnostic and treatment systems based on human-machine collaboration, and promoting widespread AI applications in large-scale lung cancer screening, personalized therapy, and comprehensive disease management. These efforts aim to contribute to the precise prevention and treatment of lung cancer.