S3021
Invited Speaker
ESTRO 2026
actionable anatomical change from computational noise across all institutions. Each centre must establish locally validated frameworks and action levels, characterise the accuracy of its own CBCT dose calculation pipeline, and interpret dosimetric differences within its specific uncertainty budget, providing the necessary foundation for the decision frameworks and traffic-light systems discussed in subsequent talks in this session. 5378 To adapt or not adapt: TLP versus daily dose calculation on CBCT Dylan Callens Laboratory of Experimental Radiotherapy, KU Leuven, Leuven, Belgium. Department of Radiation Oncology, UZ Leuven, Leuven, Belgium A traffic light protocol (TLP) serves two main purposes in offline adaptive radiotherapy: it standardises the image registration process and provides a structured, colour coded communication tool between professional groups. Fraction level communication is triggered by orange or red flags and may prompt further review or plan adaptation. While anatomy based TLPs are valuable for harmonising registration practice, they have inherent limitations as communication and decision support tools. In particular, they rely on visual anatomical assessment and do not convey dosimetric information. As a result, it remains unclear to what extent a reported anatomical change is associated with a meaningful dosimetric effect, which introduces ambiguity into decisions on whether treatment adaptation is warranted. Recent advances in CBCT based dose calculation and automation now make it possible to address this uncertainty more systematically through dose monitoring. In this study, twenty two patients with locally advanced non-small-cell lung cancer (LA NSCLC) enrolled in the prospective adaptive radiotherapy trial ECLAIR (NCT07259447) were included for parallel evaluation of anatomy based TLP reporting and fraction specific dose monitoring. Against this background, this presentation discusses daily CBCT-based dose monitoring in patients with LA- NSCLC in relation to an anatomy-based traffic light protocol. It describes how dose monitoring was automated into a level-3 end-to-end workflow, reflects on the advantages and limitations of this approach, and demonstrates what can be learned from quantifying daily delivered dose and its temporal patterns throughout treatment. Finally, the association between anatomical TLP flagging and constraint- specific delivered dose is revealed using linear mixed- effects modelling.
difficulties in defining recommendations. A well- established terminology and thorough understanding of the state-of-the-art are prerequisites to favouring a clear consensus. During this introductory talk, we will define basic terms for dose to medium and dose to water including refinements for dose to medium in medium and dose to water in medium. This will be presented in a simplified manner sufficient for setting the scene. We will also describe several available dose calculation algorithms including most commercial ones as it is not always clear which quantity is reported by a given algorithm. The focus of the talk will be high-energy X-rays external beam radiotherapy but some information specific to charged particle therapy will also be provided. 5377 Dose calculations on CBCT: Technical aspects Ciaran Malone PhD program, Erasmus University, Rotterdam, Netherlands If the CBCT recalculation shows a ~1.5 Gy difference to an OAR or target, should we adapt? This talk examines whether such differences represent actionable clinical signals or fall within the cumulative noise inherent to the CBCT dose calculation chain. We frame the distinction between uncertainty (irreducible spread to quantify and accept) and systematic error (correctable offset to identify and eliminate). Each step in the CBCT dose chain contributes ~1–5% uncertainty, and these contributions compound across the chain in a system- dependent manner. Before evaluating any CBCT-derived dose number, we acknowledge the uncertainty budget already accepted in conventional radiotherapy. ICRU Report 24 established that overall dose delivery accuracy should be within ±5%, and the IAEA recommends keeping random uncertainty below 5% for routine practice. A 1-2 Gy difference on a 50 Gy prescription represents 2- 4%, potentially within the noise floor we already tolerate. Added to this are inter-observer contouring variabilities, deformable image registration uncertainties, and the fundamental limitation that both the planning CT and the CBCT are single snapshots of a dynamic anatomy. Against this backdrop, we examine the technical factors specific to CBCT dose computation that individually introduce further uncertainty. These include HU-to-electron density calibration, dose calculation grid resolution, algorithm-dependent differences, the resolution mismatch between CT image, contour, and dose grids, and synthetic CT generation methods. We conclude that a universal dose-difference threshold may not be suitable to reliably separate
Made with FlippingBook - Share PDF online