S1759
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
Results:
Med Biol. May 11 2023;68(10)doi:10.1088/1361- 6560/accdb24.Pang B, Si H, Liu M, et al. Comparison and evaluation of different deep learning models of synthetic CT generation from CBCT for nasopharynx cancer adaptive proton therapy. Med Phys. Nov 2023;50(11):6920-6930. doi:10.1002/mp.16777 Keywords: online adaptive proton therapy ， deep learning
Digital Poster 3804 Skipping the Planning CT: Feasibility and
Dosimetric Safety of Direct-to-Unit Proton Therapy Lucrezia Fendillo 1,2 , Lisa Stefanie Fankhauser 1,3 , Eliane Garlock 1,4 , Muheng Li 1,3 , Antony John Lomax 1,3 , Francesca Albertini 1 1 Center for Proton Therapy, Paul Scherrer Institute, Villigen, Switzerland. 2 Department of Physics and Astronomy, University of Bologna, Bologna, Italy. 3 Department of Physics, ETH Zürich, Zürich, Switzerland. 4 Department of Radiation Oncology, Virginia Commonwealth University Health System, Richmond, USA Purpose/Objective: Direct-to-unit (DTU) radiotherapy, in which treatment planning is initiated directly on diagnostic imaging without a dedicated planning CT (pCT), has shown promise in photon therapy to shorten time-to- treatment and reduce patient travel. Proton therapy (PT) could benefit even more because centers are scarce, often distant and operate with tightly scheduled resources. Protons however are more sensitive to tissue density and anatomical changes. It is therefore uncertain whether plans optimized on diagnostic images can be safely adapted for the first treatment day. With online adaptive PT able to compensate day-of-treatment differences, we hypothesize that, within an online adaptive workflow, re-optimizing diagnostic-template plan on the first-day CT will match standard pCT-based planning – enabling omission of a dedicated pCT. Material/Methods: We retrospectively analyzed ten patients (three abdominal, two head-and-neck, five brain), for which diagnostic CTs or MRIs (converted to synthetic CTs using a deep-learning model [1]) were acquired 6 days to 19 months before the clinical pCT (Figure 1). The clinical plan optimized on the pCT served as the gold standard. To emulate a DTU workflow, clinical contours were propagated from the pCT to the diagnostic image and reviewed/edited by a radiation oncologist. A plan (template plan) was generated on the diagnostic image knowing only (i) the beam/field directions used clinically and (ii) the clinical goals to be met,— i.e. optimization objectives were defined
The dose distribution calculated on the deformed planning CT was used as the ground truth. The gamma passing rates (2%, 2 mm) of the sCT planned dose reached 99.42 ± 0.48% and 99.36 ± 0.45% for the prostate and brain cases, respectively. The values decreased to 94.59 ± 2.00% and 93.42 ± 1.11% for the predicted doses, but subsequently improved to 95.94 ± 0.48% and 94.40 ± 1.39% for the weight-predicted doses. In terms of absolute differences in target D98, the dose prediction step showed smaller deviations than the weight prediction step, indicating that it still provided accurate target dose estimation and thus laid a solid foundation for reliable weight prediction. The entire automated workflow excluding the structure contouring step can be completed in less than one minute. Conclusion: This study, for the first time, integrates deep learning models to establish a fully automated online adaptive proton therapy workflow. The workflow demonstrates satisfactory results with a remarkably short execution time, and the stepwise evaluation of each component provides valuable insights for future improvement and development. References: 1.Mohan R, Grosshans D. Proton therapy - Present and future. Advanced drug delivery reviews. Jan 15 2017;109:26-44. doi:10.1016/j.addr.2016.11.0062.Paganetti H, Botas P, Sharp GC, Winey B. Adaptive proton therapy. Phys Med Biol. Nov 15 2021;66(22)doi:10.1088/1361- 6560/ac344f3.Qiu Z, Olberg S, den Hertog D, Ajdari A, Bortfeld T, Pursley J. Online adaptive planning methods for intensity-modulated radiotherapy. Phys
Made with FlippingBook - Share PDF online