ESTRO 2025 - Abstract Book

S120

Invited Speaker

ESTRO 2025

Finally, I will present some additional metrics and evaluations that should be considered when assessing AI foundation models for clinical and clinical research use.

Learning objectives: a. Understand what foundation models are and how they are different from task-specific models b. Understand how to train foundation models, how to assess their effectiveness. c. Metrics for evaluation of AI and AI foundation models.

4847

Speaker Abstracts Revisiting reference dosimetry Anna Subiel Radiotherapy and Radiation Dosimetry, National Physical Laboratory, Teddington, United Kingdom. Medical Physics and Biomedical Engineering, University College London, London, United Kingdom Abstract: Radiotherapy with ultra-high dose rates (UHDR), commonly referred to as FLASH radiotherapy, holds significant promise for minimizing side effects in cancer treatment. The ability to deliver highly concentrated doses in a very short period could revolutionize clinical outcomes. However, for FLASH radiotherapy to reach its full potential, precise dosimetry of UHDR beams is crucial. Accurate dose measurements ensure that the therapeutic benefit is maximized while minimizing harm to healthy tissue. Reference dosimetry plays a critical role in this process. It is a procedure carried out within a user's specific beam configuration to establish a traceable dose measurement at a defined point in a standard geometry. The procedure typically relies on a well-defined calibration protocol, a detector with a known calibration factor, and the application of appropriate correction factors. Each of these elements requires careful consideration to ensure accurate reference calibration for UHDR beams. Current clinical calibration protocols for reference dosimetry predominantly rely on ionization chambers, which have intermediate to large volumes (ranging from 0.1 cm³ to 0.6 cm³). These reference-class chambers are expected to remain the gold standard for dosimetry in the foreseeable future. However, their use in FLASH radiotherapy presents unique challenges. The ultra-high dose rates used in FLASH therapy necessitate significant corrections to the detector readings, as traditional ionization chambers are not designed to measure such rapid dose deliveries with high accuracy. In this talk, I will explore the key challenges and considerations involved in achieving high-precision reference dosimetry for both National Metrology Institutes (NMIs) and end users (clinical physicists) working with UHDR beams.

4848

Speaker Abstracts Status of the instrumentation Francesco Romano Catania Division, National Institute for Nuclear Physics, Catania, Italy

Abstract:

FLASH radiotherapy (RT) is attracting a increasing interest since the first investigations carried out in 2014. Several preclinical studies worldwide have demonstrated that ultra-high dose rate (UHDR) beams provide an improvement of normal tissue sparing, compared to conventional RT, while maintaining same tumour control probability (FLASH effect). However, to fully understand the mechanisms behind the effect and to support the clinical translation of