S3007
Invited Speaker
ESTRO 2026
7. Kolkman-Deurloo I.K. et al., Brachytherapy, 23(6), PHSOR07, pp. S61, 2024.
combined intracavitary-interstitial applicators, and the implementation of individualized multiparametric dose-prescription concepts [1]. The introduction of IGABT, however, has substantially increased both procedural time and complexity. A time-action analysis involving 56 patients demonstrated that the current median duration from patient arrival to discharge for a single-fraction application exceeds eight hours [2]. This duration further increases when multiple fractions are administered within the same application – typically with an interval of at least six hours – often necessitating an overnight hospital stay with the applicator and needles remaining in situ. Such prolonged immobilization imposes a considerable
5334 Artificial intelligence can accelerate the brachytherapy treatment processes Bruno Fionda
UOC Degenze di Radioterapia Oncologica, Fondazione Policlinico Universitario Agostino Gemelli IRCCS, Rome, Italy Artificial intelligence (AI) is increasingly recognised as a valuable tool to accelerate workflows in interventional radioterapy (modern brachytherapy) by improving efficiency, consistency, and standardisation across all stages of the treatment process. In current clinical practice, this modality involves several complex and time-consuming steps, including patient consultation, imaging, contouring, applicator reconstruction, treatment planning, delivery, and follow-up. AI-based approaches, particularly those based on machine learning and deep learning, have shown the potential to significantly reduce the time required for many of these tasks while maintaining clinically acceptable accuracy. AI plays a central role in the development of decision support systems, enabling the integration of large datasets, advanced imaging analysis, and predictive modelling. These systems can assist clinicians in treatment selection, personalise therapeutic strategies, and support decision-making by incorporating multiple variables, including patient characteristics and clinical context. At the same time, AI facilitates repetitive and labour-intensive tasks, allowing clinicians to focus more on patient care and complex decision processes . One of the most advanced applications is automated segmentation, where deep learning models have demonstrated strong performance in delineating targets and organs at risk, reducing inter-observer variability and improving workflow efficiency. Similarly, AI-driven approaches to treatment planning, including knowledge-based and optimisation algorithms, can reduce the number of manual iterations required and improve plan quality. Additional applications include applicator selection and positioning, optimisation of source dwell positions, and motion management, all contributing to more precise and efficient treatment delivery . There is growing evidence in the scientific literature supporting these applications; however, their actual use in routine clinical practice remains limited. This reflects a gap between technological development and real-world implementation, where challenges such as data availability, validation, integration into clinical
mental and physical burden on patients [2]. Moreover, the sustainability of current IGABT
workflows is increasingly uncertain given the projected rise in cancer incidence and the worsening shortages in the radiotherapy workforce, which is already functioning under substantial operational strain. As outlined in [3], one of the key prerequisites for ensuring future accessibility of radiotherapy is to reduce the average workload per patient, thereby enabling more patients to be treated per radiotherapy professional without compromising job satisfaction. Applied to LACC IGABT, this necessitates the development of new technologies and clinical processes that accelerate workflow steps and minimize hands on-time where feasible. Given that treatment planning represents the most time-consuming component of the LACC IGABT workflow [2], several strategies for treatment planning workflow optimisation can be envisioned. This presentation will focus on the validation of automated implant geometry assessment using electromagnetic tracking (EMT) to enable automated applicator and needle reconstruction, as previously proposed in [4,5], and on the development of automated multi-criteria optimization for LACC IGABT. Our in-house developed medical device software, BiCycle, currently represents the only fully automated treatment planning solution integrated into a standard clinical workflow for LACC IGABT outside a research environment [6,7]. The roadmap toward its clinical introduction, in compliance with the requirements of the EU Medical Device Regulation (EU-MDR) will be outlined. References: 1. Tanderup K. et al., Radiother Oncol 120: 365-369, 2016. 2. van Vliet-Pérez S. et al., Brachytherapy 23:274-281, 2024. 3. Petit S. et al., Radiother Oncol 211:111-075, 2025. 4. Gomez-Sarmiento I. et al., Med Phys 51(2):799-808, 2024. 5. Heerden van L. et al., Radiother Oncol 154:249-254, 2021. 6. Rossi L. et al., Radiother Oncol 210:111029, 2025.
Made with FlippingBook - Share PDF online