AI in Lung Cancer Screening: Summary from the RAIN Network Meeting

From Pilot Project to National Perspective

How can AI strengthen the early detection of lung cancer in Denmark? That question brought together more than 60 participants at Health Innovation Centre of Southern Denmark in Odense on September 30, where RAIN (AI in Healthcare) hosted a network meeting on AI in lung cancer screening. 

The day focused on how AI is already being used in pilot studies that may pave the way for a national screening program. Presentations by researchers, clinicians, and public authorities provided insight into the experiences from the PLUS Project, the development of the AI-RAPTOR algorithm, and the interdisciplinary collaborations enabling the integration of the technology into clinical practice. 

AI-RAPTOR: From Idea to Clinical Application

The network meeting opened with a presentation on AI-RAPTOR by Frederik Duedahl and Simon Lyck Bjært Sørensen, PhD students at the Department of Cardiothoracic Surgery at Odense University Hospital. AI-RAPTOR is an annotation platform and algorithm for identifying cancerous nodules in CT scans. 

They showcased how the RAPTOR platform enables precise marking and analysis of large volumes of image data, built on more than 20,000 CT scans and extensive radiological data. Their presentation highlighted RAPTOR’s potential not only as a decision-support tool in clinical practice but also as a training and research resource for developing new AI models. 

PLUS Project: Pilot Study in the Region of Southern Denmark

Next, Michael Stenger, senior consultant at the Department of Cardiothoracic Surgery at Odense University Hospital and project lead for PLUS, presented the status and findings of the ongoing pilot study on lung cancer screening in Region of Southern Denmark. 

The study involves 1,000 citizens aged 60-74 with a high risk of lung cancer and is conducted using annual low-dose CT scans and supplementary lung function tests. 

Michael Stenger highlighted both clinical results and organizational insights, including: 

• The potential of the screenings to reduce mortality.
• Capacity challenges, including the urgent need for more thoracic radiologists.
• The importance of AI in supporting the assessment of numerous scans.
• Perspectives on managing incidental findings and communicating with citizens.
• The presentation emphasized that AI cannot stand alone; it must function as an integral part of the clinicians’ work. 

National Context: The Danish Health Authority's Pilot Study

Kristoffer Lande Andersen from the Danish Health Authority discussed the PLUS Project within a national framework. He explained why the Danish Health Authority decided in 2023 to launch a pilot study before determining whether a full-scale national screening program could be realized. 

The pilot aims to gather insights on: 

• How best to identify and invite citizens to screening.
• Resource consumption and the ripple effects on other patient groups.
• Organizational and logistical aspects of a national screening program.
• Utilizing AI as a tool to address bottlenecks.

AI’s Broader Potential: From Lung Cancer to Aortic Aneurysms

Mads Liisberg, physician and postdoc at the Research Unit for Cardiothoracic Surgery at Odense University Hospital, presented how the experiences from AI-RAPTOR could be extended to other areas, such as personalized risk assessment for aortic aneurysms. 

With AI models capable of analysing CT scans and patient data, it becomes possible to create individualized risk profiles, reducing unnecessary surgeries while identifying overlooked patients. His presentation demonstrated how the technology enables a shift from a "one-size-fits-all" approach to more tailored risk evaluations. 

Dialogue between Experts

Participants were invited to engage in dialogue-driven breakout sessions, interacting with presenters - including Mads Liisberg, Frederik Duedahl, and Simon Lyck Bjært Sørensen, as well as several support and expert units from Region of Southern Denmark: 

• AI Competence Unit with Nicklas Hauge Hansen and Kathrine Holtz Paarup: Advising on the process from idea to operation.
• Information Security and Compliance in Regional IT with Kirstine Louise Jensen, Lars Henrik Steenfeldt, and Thomas Arvid Gadeberg Bentsen: Guidelines and risk assessment in relation to MDR and AI Act.
• OPEN’s GCP Unit with Jeanett Høg Jørgensen and Kirsten Klejs Breiting: Supporting clinical AI trials.
• CAI-X – Danish Centre for Clinical Artificial Intelligence at Odense University Hospital and University of Southern Denmark, with Maiken Wolderslund, Peter Børker Nielsen, and Benjamin Rasmussen: Research and innovation in clinical AI.

The groups discussed topics such as development processes, regulatory frameworks, information security, data availability, and the practical steps needed to bring AI solutions into clinical practice.

Final Perspectives from the Day

A collective wrap-up closed the network meeting. Discussions emphasized that AI is key to realizing a national lung cancer screening program, but the development requires close collaboration between clinicians, researchers, public authorities, and technical experts. 

Joint discussions also underscored an important perspective: AI is not the answer to all challenges in healthcare. Many issues can be addressed through more simple technologies. The point was that development should always start from the need, not the technology - and AI only makes sense where it creates real value for patients, staff, and organizations. 

The day clearly demonstrated the value of bringing diverse expertise together within the RAIN network to share knowledge, foster dialogue, and inspire the use of AI in the healthcare system. 

Thank you to all presenters and participants for an engaging and inspiring day on the future of lung cancer screening with AI!