ESTRO 2025 - Abstract Book

S3021

Physics - Image acquisition and processing

ESTRO 2025

Conclusion: Deterministic behavior for the deep learning sCT algorithm was observed. Our study indicates the impact of repositioning and time gaps on sCT constancy and reproducibility, which limits the feasibility of using volunteer scans as a QA method but also shows the level of positioning uncertainty within a mask. To investigate long-term changes, the impact on dose distribution, and the effects of anatomical and positioning variances, the study is planned to be conducted over a total of 52 weeks (currently at week 5). To assess sCT generation constancy without the variability introduced by human subjects, developing and implementing dedicated phantoms for end-to-end testing is urgently required.

Keywords: Synthetic CT (sCT), Quality Assurance (QA)

References: 1. Masitho, S., et al., Accuracy of MRI-CT registration in brain stereotactic radiotherapy: Impact of MRI acquisition setup and registration method. Zeitschrift für Medizinische Physik, 2022. 32(4): p. 477-487. 2. Villegas, F., et al., Challenges and opportunities in the development and clinical implementation of artificial intelligence based synthetic computed tomography for magnetic resonance only radiotherapy. Radiotherapy and Oncology, 2024. 198: 110387.