A study by researchers from the University of Liverpool and the Research Institute for Diagnostic Accuracy, Netherlands, has demonstrated that artificial intelligence (AI) can significantly improve the efficiency of lung cancer screening.
Published in the European Journal of Cancer, the study reveals that AI can accurately rule out negative low-dose CT (LDCT) scans, potentially reducing the workload of radiologists by up to 79%.
Lung cancer affects more than 48,000 people in the UK every year, and early detection is crucial for improving survival rates. The UK Lung Cancer Screening (UKLS) trial has already shown that LDCT screening can save lives by detecting lung cancer in high-risk individuals before symptoms appear.
In this latest study, researchers tested an AI tool developed by Coreline Soft, Co Ltd., South Korea, using UKLS trial data. The AI successfully identified scans without significant lung nodules-representing the majority of cases-even among high-risk individuals. This allows radiologists to focus their expertise on cases that require further analysis, improving efficiency while maintaining accuracy in lung cancer detection.
A key finding of the study is that all confirmed lung cancer cases were among the scans flagged by the AI for further review. This ensures that no cancers were missed while significantly reducing the number of scans requiring manual assessment. The study's success was made possible by the high-quality radiology reporting from the UKLS trial and long-term follow-up data, which provided a reliable dataset for AI validation.
Professor John Field, lead author and Professor of Molecular Oncology at the University of Liverpool, emphasised the study's importance: "Implementing low-dose CT screening for lung cancer is highly beneficial, but it comes with logistical and financial challenges. Our research suggests that AI could play a crucial role in making screening programs more efficient while maintaining diagnostic confidence."
Co-lead author Professor Matthijs Oudkerk, Professor Emeritus of Radiology at the University of Groningen, Chief Scientific Officer of the Institute for Diagnostic Accuracy added: "This is the first chest AI validation study performed in a real-world consecutive lung cancer screening program, with histological proven outcomes of lung cancer and a more than 5-years follow-up for disease free survival. Therefore, a milestone for further AI validation in terms of methodology and accuracy with results that can be translated to medical implementation."
Lung cancer screening programs are expanding worldwide, and AI-driven tools like the one tested in this study has the potential to be instrumental in optimising healthcare resources, reducing costs, and ensuring timely diagnoses. Further research and validation studies will help refine these AI models.
The paper, 'Histological Proven AI Performance in the UKLS CT Lung Cancer Screening Study: Potential for Workload Reduction', was published in the European Journal of Cancer (DOI:10.1016/j.ejca.2025.115324).
