IAEA study shows AI boosts cancer contouring quality and speeds radiotherapy planning

To investigate AI’s real-world benefits, the IAEA focused on head and neck cancer — one of the most challenging areas for contouring due to complex anatomy.

A landmark international study coordinated by the International Atomic Energy Agency (IAEA) has confirmed that artificial intelligence (AI) can enhance both the quality and efficiency of contouring organs at risk — a critical and often time-consuming step in cancer radiotherapy planning. With participation from 23 countries across multiple regions, the ELAISA Study provides the strongest global evidence to date that AI-assisted contouring can relieve workforce shortages and expand access to life-saving cancer treatment, especially in low- and middle-income countries (LMICs).

AI Strengthening a Crucial Step in Radiotherapy

Contouring — outlining tumours and surrounding organs-at-risk — is vital for delivering safe, precise radiotherapy. Yet this process varies widely between clinicians, causing inter-observer variability that can affect treatment accuracy. Historically, instructor-led workshops helped improve consistency, but demand for radiotherapy continues to outpace capacity.

Nearly half of all cancer patients will need radiotherapy at some point, yet access remains deeply unequal. The IAEA-led Lancet Oncology Commission on Radiotherapy and Theranostics projects a staggering global shortfall: over 84,000 additional radiation oncologists will be required by 2050 to meet rising cancer needs.

“As cancer cases and treatment complexity increase, radiation oncologists will have to spend even more of their already limited capacity on contouring,” said Dr May Abdel-Wahab, Director of the IAEA Division of Human Health. “AI can become an essential tool to support this workload.”

Testing AI in Real-World Clinical Settings

To investigate AI’s real-world benefits, the IAEA focused on head and neck cancer — one of the most challenging areas for contouring due to complex anatomy.

Almost 100 radiation oncologists from 22 radiotherapy centres in LMICs took part from countries including Albania, Argentina, Bangladesh, India, Indonesia, Kenya, Nepal, Pakistan, Tunisia and Uganda. Aarhus University Hospital in Denmark contributed 16 head and neck cancer cases to support the study.

Participants were divided into two groups: one using AI-assisted contouring and another using manual methods. After an IAEA-led workshop on AI tools, all participants performed additional contouring rounds — both immediately after training and six months later — allowing researchers to measure short- and long-term effects.

AI Assistance Improved Quality and Saved Time

Results demonstrate clear, measurable benefits:

• AI significantly reduced inter-observer variability, improving consistency and quality.

• Contouring time decreased substantially with AI — even without prior training.

• Instructor-led teaching further magnified the time-saving effect, though it only improved quality for two specific organs-at-risk.

• Follow-up rounds showed lasting improvements, indicating that clinicians retain efficiency and confidence with repeated AI use.

“The ELAISA Study shows that teaching combined with AI-assisted contouring was the most effective strategy to reduce contouring time,” said Professor Jesper Grau Eriksen, one of the lead investigators. “If applied appropriately, AI-assisted contouring tools can save valuable resources and enable more radiation oncologists — especially in LMICs — to treat more patients.”

A Pathway to More Equitable Cancer Care

The study’s findings have major implications for global health. LMICs face both rising cancer burdens and severe shortages of oncologists, physicists and radiotherapy technicians. By reducing workload and improving quality, AI can help make radiotherapy more accessible, equitable and efficient worldwide.

The results have been published in the Journal of Global Oncology and presented at leading conferences, including the annual meetings of the European Society for Radiotherapy and Oncology (ESTRO).

As cancer cases continue to rise globally, AI-assisted tools — combined with targeted training and international collaboration — offer a promising avenue for expanding radiotherapy access and improving patient outcomes.