IAEA Launches Global AI Research Project to Improve Cancer Radiotherapy

The International Atomic Energy Agency (IAEA) has launched a groundbreaking two-year coordinated research project (CRP) to explore how artificial intelligence (AI) can enhance one of the most critical steps in cancer radiotherapy—contouring. This process involves outlining the tumour and surrounding tissues to precisely target radiation therapy, and errors or inconsistencies can lead to under-treatment of cancer or damage to healthy tissue.

The new project will not only assess how AI-based tools can reduce variation and bias among clinicians but also examine how explainable AI (XAI) can enhance medical understanding and decision-making in complex radiotherapy cases.

The Need for Precision in Contouring

In radiotherapy, “contouring” defines three major components:

• Gross Tumour Volume (GTV) – the visible and image-detectable tumour

• Clinical Target Volume (CTV) – the surrounding tissues that may harbour microscopic cancer cells

• Organs-at-Risk (OAR) – nearby healthy tissues that must be spared from unnecessary radiation

Despite improvements in imaging technologies and treatment techniques, contouring still involves significant subjectivity and variability among oncologists. This variation can lead to inconsistent outcomes, as Tomoaki Tamaki, Head of Applied Radiation Biology and Radiotherapy at the IAEA, explained:

“Missteps during contouring can significantly impact treatment outcomes — be it from missing the tumours or irradiating normal tissues unnecessarily. Limitations in diagnostic imaging, difficulties in identifying tumours accurately, and the increasing global cancer burden can all add to the contouring challenges clinicians face.”

Building on Prior Success

This CRP builds upon the IAEA’s previous 2024 research project, which assessed AI’s value in identifying organs-at-risk. That project demonstrated that AI, particularly deep learning, can significantly reduce the time clinicians spend on delineating healthy anatomy and enhance consistency between practitioners. The new project expands on that foundation, aiming to automate and explain the more nuanced task of identifying gross tumour volume, especially for head and neck cancers, which are anatomically complex.

AI and XAI: A Dual-Focus Approach

AI has already shown promise in the form of auto-segmentation algorithms that reduce workload and improve consistency. However, questions remain about bias, transparency, and trust in AI-generated contours. That’s where Explainable AI (XAI) enters the scene.

Explainable AI aims to demystify the decision-making process of deep learning models, offering clinicians insights into why the AI has recommended certain contours—vital for gaining trust and ensuring the technology aligns with expert medical judgment.

The CRP is structured in three phases:

1. Phase One will compare manual contouring with AI-assisted methods, both with and without XAI.

2. Phase Two will investigate whether XAI reduces observer bias in interpreting AI outputs.

3. Optional Phase Three will explore optimal methods for visualizing and presenting AI explanations to clinicians.

Participants will also attend IAEA-led educational sessions on AI in radiotherapy and best practices in contouring.

Research Objectives and Global Participation

The primary goal of the CRP is to improve the quality and accuracy of radiotherapy for head and neck cancers, which affect over 900,000 people globally each year. The project seeks to evaluate whether combining e-learning, AI-assisted contouring, and XAI tools can significantly reduce inter-observer variation—the discrepancies between different oncologists’ delineations of the same tumour.

The study is open to all IAEA Member States, and participating institutions must meet the following eligibility criteria:

• Treat at least 20 head and neck cancer patients per year

• Use 3D radiotherapy with CT-based planning

• Maintain a robust internet connection for case data exchange and virtual workshops

• Have a team of at least 3 to 4 oncologists routinely involved in head and neck cancer treatment

Interested institutions must submit a Proposal for Research Contract or Agreement by 30 August 2025, using the designated template available on the IAEA’s Contract Research Administration (CRA) portal. Proposals should prioritize the involvement of women and young researchers, in line with the IAEA’s commitment to equity and capacity building.

A Step Forward in Global Cancer Care

As the global cancer burden rises—particularly in low- and middle-income countries—the need for consistent, high-quality cancer treatment becomes increasingly urgent. By reducing contouring inconsistencies, AI can help ensure more equitable care, especially in resource-limited settings where oncologists are overburdened.

This CRP will not only generate new evidence on the practical value of AI and XAI in clinical settings but also help develop a new cadre of professionals skilled in using AI ethically and effectively in radiation oncology.

With AI promising to transform the precision of cancer care, this IAEA initiative represents a crucial step in aligning global technology trends with real-world clinical practice.