Translation. Region: Russian Federation –
Source: International Atomic Energy Agency –
An important disclaimer is at the bottom of this article.
Home
News
IAEA study finds benefits of AI-based contouring for cancer patients
An IAEA coordinated research project has demonstrated how AI can expand access to radiotherapy worldwide.
December 19, 2025
Peter Lee, IAEA Department of Nuclear Sciences and Applications
Lisbeth Cordero Mendes, IAEA Department of Nuclear Sciences and Applications
A radiation oncologist contours a patient's head and neck tumors. Photo: G. Ferraris
A 23-country study demonstrates the safety and benefits of using artificial intelligence in a key and often most labor-intensive step of the cancer treatment process: mapping organs at risk. By adding unique data from low- and middle-income countries to the growing body of scientific evidence, the IAEA Coordinated Research Project (ELAISA study) shows how this technology can expand access to radiotherapy worldwide.
Contouring tumors and surrounding healthy tissue (organs at risk) is essential for the optimal, safe, and effective use of radiation therapy for cancer treatment. However, differences in how different specialists perform contouring (i.e., interobserver variability) can impact both the accuracy and consistency of radiation therapy planning. Previous studies have suggested that instructor-led training workshops can reduce interobserver variability.
Nearly half of cancer patients require radiation therapy at some point, yet access to this type of treatment is underutilized worldwide, in part due to a shortage of clinically trained specialists. According to Lancet Oncology Commission on Radiation Therapy and Theranostics, led by the IAEA, more than 84,000 radiation oncologists will be needed to meet the global demand for cancer treatment, which will be 35.2 million new cases by 2050. “This figure includes an increase of more than 60% in the number of radiation oncologists in 2022,” says the director of IAEA Division of Human Health and commission co-chair May Abdel-Wahab. "As cancer incidence and treatment complexity increase, radiation oncologists will have to dedicate even more time to delineating cancerous tissue and surrounding healthy tissue in settings where their capabilities are already limited."
AI is being considered as an adjunct in the treatment of head and neck cancer
To address these challenges in radiation oncology, the IAEA has explored how artificial intelligence (AI) can assist in head and neck cancer delineation in low- and middle-income countries (LMICs).
AI-based algorithms have shown promising results in automatic structure delineation (autosegmentation), but this has primarily been observed in retrospective studies. Until recently, the actual clinical benefit in the context of LMICs and in terms of interobserver variability remained largely unexplored.
"Using AI to assist with contouring could be an important tool to improve the efficiency of radiation oncologists," Abdel-Wahab notes.
With the participation of radiation oncologists from 22 countries
The IAEA study involved approximately 100 radiation oncologists from 22 different radiotherapy centres in Azerbaijan, Albania, Argentina, Bangladesh, Belarus, Costa Rica, Georgia, India, Indonesia, Jordan, Kazakhstan, Kenya, Kyrgyzstan, Malaysia, Moldova, Mongolia, Nepal, Pakistan, North Macedonia, Sudan, Tunisia and Uganda, while Aarhus University Hospital in Denmark provided data on 16 head and neck cancer cases.
During the study, radiation oncologists were randomly divided into two groups: one group used AI to delineate organs at risk, while the other used manual methods. Following an IAEA online workshop on AI-assisted delineation, both groups continued tumor delineation, first using their original approach and then using AI. Six months later, the final phase of the study, using AI, took place.
Improving the quality of contouring using AI
Results from the IAEA Coordinated Research Project demonstrated that AI can not only improve the quality of contouring by significantly reducing interobserver variability, but also reduce the time it takes to complete the contouring process, even without prior instruction. Instruction only improved the quality of contouring for two at-risk organs, but it significantly enhanced the time savings associated with AI-based contouring. This effect was also observed over time during short- and long-term studies of the instructor-led workshop.
“ELAISA study "This study shows that training combined with AI-assisted contouring was the most effective strategy for reducing the time it takes to contour," explains Jesper Grau Eriksen, Professor of Clinical Medicine at Aarhus University and one of the lead researchers on the project. "When used appropriately, the safe implementation of AI-based contouring tools could save resources and enable more radiation oncologists, particularly those in LMIC settings, to care for even more patients."
The results of the study were published in the journal "Global Oncology" and presented at annual meetings European Society of Radiotherapy and Oncology.
Related news
Related resources
Read more
Please note: This information is raw content obtained directly from the source. It represents an accurate account of the source's assertions and does not necessarily reflect the position of MIL-OSI or its clients.