S2017
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
Material/Methods: Ten previously treated patients (vertebrae, femur, ribs, sternum) prescribed 8 Gy in one fraction were retrospectively re-planned. For each case, an automatic CTV was generated on the involved bone and expanded to a PTV using physician-approved margins. An IMRT AutomaticPlan was created via automated segmentation and fluence optimization and compared with the original ClinicalPlan (IMRT or VMAT). Geometric agreement between automatic and clinical CTVs was quantified with Dice index and centroid shift. Plan quality metrics included a target- based conformity index, CI , homogeneity index (HI), gradient index (GI), mean gradient (GM), and hot spots inside/outside the PTV. Deliverability was characterized by MLC duty cycle and a vendor- independent complexity score. Physics QA used Monte Carlo re-calculation and gamma pass rate for the full dose matrix. Paired t-tests or Wilcoxon signed-rank tests were applied ( α = 0.05). Results: Automatic CTVs showed good geometric fidelity (mean Dice 0.79, range 0.57–0.96; mean centroid shift 0.40 cm, range 0.04–0.92 cm). The lowest Dice values occurred for ribs and targets at the cervico-thoracic junction (C6–D2), whereas mid-cervical segments (e.g., C3–C5) showed high agreement. CI was slightly lower for AutomaticPlan than ClinicalPlan (0.50 vs 0.62; p = 0.19). HI improved significantly with automation (difference − 0.041; p = 0.0001), indicating greater dose homogeneity. No significant differences were observed for GI or GM. Hot spots were reduced for AutomaticPlans both inside and outside the PTV (p ≈ 0.003–0.004). Complexity score was significantly lower for AutomaticPlans (0.053 vs 0.118; p = 0.016), while MLC duty cycle was similar (p = 0.15). Monte Carlo gamma pass rates for the full dose matrix were high and comparable between plan types (mean 99.1%, all ≥ 96.5%). These findings support the operational feasibility of a ~1-hour pathway.
Conclusion: This study has successfully proposed a streamlined large animal SBRT workflow and adapted to porcine STAR, with acceptable coverage and OAR sparing, potentially lowering logistical and cost barriers that inhibit these studies. References: 1. Joshi K, Katam T, Hegde A, Cheng J, Prather RS, Whitworth K, et al. Pigs: Large animal preclinical cancer models. World J Oncol. 2024 Apr;15(2):149– 68. 2. Knutson NC, Samson PP, Hugo GD, Goddu SM, Reynoso FJ, Kavanaugh JA, et al. Radiation therapy workflow and dosimetric analysis from a phase 1/2 trial of noninvasive cardiac radioablation for ventricular tachycardia. Int J Radiat Oncol Biol Phys. 2019 Aug 1;104(5):1114–23. Keywords: Animal, cardiac, arrhythmia
Digital Poster 4929
CT-to-RT in one hour: feasibility of automatic planning for palliative bone radiotherapy Anna Sardo 1 , Nicola Padula 1 , Giorgio Bosticardo 1 , Gianni Intermite 1 , Simone Baroni 2 , Rachele Petrucci 2 , Alessia Reali 2 , Claudio Scaffidi 2 , Francesco Lucio 1
1 Medical Physics, ASLCN2, Verduno, Italy. 2 Radiotherapy, ASLCN2, Verduno, Italy
Purpose/Objective: To validate a clinically deployable automated workflow for single-fraction (8 Gy) palliative bone radiotherapy by demonstrating equivalence in core dosimetric endpoints and preserved physics QA versus clinician- generated plans, enabling a rapid (~1-hour) CT-to- treatment pathway.
Made with FlippingBook - Share PDF online