AI for biology and physics in radiotherapy planning and adaptation: a joint Physics-Biology workshop

Chairs:

• Bartek Tomasik
• Heidi Lyng

 

Motivation & Description

Artificial intelligence (AI) and machine learning (ML) are rapidly transforming radiotherapy, with applications spanning imaging, treatment planning, adaptive workflows, and outcome prediction. Yet AI/ML enters radiotherapy through two fundamentally different scientific cultures. In radiotherapy physics, models are expected to operate within well-defined physical laws, deterministic behaviour, and strict safety constraints. In cancer biology, by contrast, AI/ML is used to model heterogeneous, dynamic, and probabilistic processes such as treatment response, toxicity, and tumour control. This difference creates a major translational challenge: AI/ML tools may be physically precise but biologically oversimplified, or biologically informative but difficult to validate and safely implement in clinical workflows.

This joint physics-biology workshop addresses a key question: how can biological variability be meaningfully and safely integrated into physics-based radiotherapy planning and adaptation? Participants will explore how AI/ML can connect tumour biology, biomarkers, imaging, dose calculation, optimisation, and adaptive decision-making across the radiotherapy pathway. Particular emphasis will be placed on interpretability, robustness, uncertainty quantification, validation, and patient safety.

The workshop will provide a joint forum for physicists, biologists, and clinicians to compare assumptions, constraints, and validation standards across disciplines, and to define realistic interdisciplinary AI/ML use cases for radiotherapy. This fits well with the ESTRO Physics Workshop format, which is intended as an interactive, collaboration-building environment focused on scientific exchange and tangible outcomes rather than a traditional course or congress.

 

Planned Outcomes

By the end of the workshop, participants will:

• clearly articulate the conceptual differences between AI applications in biology and in physics;
• identify where and how biological knowledge, physical laws, and clinical constraints can be integrated into AI/ML models;
• define validation, uncertainty, and safety frameworks suitable for interdisciplinary AI/ML models in radiotherapy;
• formulate clinically relevant AI/ML problem statements aligned with radiotherapy planning and adaptation;
• develop a shared interdisciplinary language between physicists, biologists, and clinicians;
• outline realistic pathways for clinical translation, including regulatory, ethical, and implementation considerations;
• generate concrete framework ideas or group-designed concepts for future collaborative work in this area.

 

Applicant Eligibility

Radiation biologist, physicists and physician scientists (with physics or biology focus) at all levels of their academic career (from students to senior scientists).