ESTRO 2025 - Abstract Book

S3612

Physics - Quality assurance and auditing

ESTRO 2025

Conclusion: This audit highlights the essential role of standardized QA and consistent CoP application for absorbed dose determination in multi-institutional radiotherapy studies. This standardization ensures accurate dosimetry measurements and treatment quality, safeguards patients, and enables meaningful comparison of outcomes across centeres.

Keywords: Clinical Trials, Dosimetry Audit, Proton+Photons

References: [a] Mortensen H, Populaire P, Hoffmann L, et al. Proton versus photon therapy for esophageal cancer - A trimodality strategy (PROTECT) NCT050555648: A multicenter international randomized phase III study of neoadjuvant proton versus photon chemoradiotherapy in locally advanced esophageal cancer. Radiother Oncol. 2023;190. doi: 10.1016/j.radonc.2023.109980. [b] Hoffmann L, Mortensen H, et al. Treatment planning comparison in the PROTECT-trial randomising proton versus photon beam therapy in oesophagela cancer: Results from eight European centres. Radiotherapy and Oncology 2022; 172. https://doi.org/10.1016/j.radonc.2022.04.029.

1376

Digital Poster A practical approach to develop in-house built medical software within the scope of the MDR Paula Bos, Herbert Beemster, Dos Moonen, Carmen Seller Oria, Rob Reurings, Ruben Bosschaert, Martijn Barsingerhorn, Peter Remeijer Department of Radiation Oncology, The Netherlands Cancer Institute, Amsterdam, Netherlands Purpose/Objective: The introduction of the Medical Device Regulation (MDR) has impacted in-house development of software severely. The extension of the definition of software as a medical device and new and stricter requirements for detailed documentation and testing make in-house development of software complex and resource-demanding. This study describes a method to simplify safe implementation of in-house built medical software within the requirements of the MDR. Material/Methods: Microservice architecture 1 was used for our applications, implementing the core functionality as multiple independent services. RabbitMQ 2 was used to orchestrate the communication among the microservices. The software runs as a pipeline triggered by DICOM data uploaded to a DICOM server, eliminating the need for a graphical user-interface. The Azure Devops 3 platform was used to build, test and deploy the applications to three different environments: test, acceptance and production. Our methodology includes automated generation of MDR compliant documentation and automated tests to assess new releases against fixed acceptance criteria, combined with visual inspection to secure performance and reliability. Development took place within a DevOps team (eight employees, 2.5 FTE), where developers and operations are jointly responsible for development, operation and support. Results: Although building the microservice framework required a considerable investment of two years of the DevOps team, currently, development and release of new functionality is very fast. Five applications have been developed within this framework. These applications use an average of three [range:1-6] microservices, where three individual microservices were shared among multiple applications. For one of the applications, consisting of four microservices, the mean(±standard deviation) deployment took 22.0±6.3, 4.5±0.5 and 4.0±1.2 minutes for the test, acceptance and production environment. Automated test results were generated in 60±13 minutes. The total time