ESTRO 2025 - Abstract Book

S2809

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2025

Conclusion: This preliminary evaluation of RapidArc Dynamic demonstrated improved hippocampal sparing and reduced exposure to critical organs in whole brain radiotherapy cases. Future research will refine the RapidPlan model for VMATp to further maximize the benefits of this technology.

Keywords: RapidArc Dynamic, Hippocampal Sparing

2294

Proffered Paper Principal component analysis statistical shape model to predict breast dose of pre-CT era radiotherapy pediatric patients in long-term outcome studies Taylor G Meyers 1 , Caleb S OConnor 2 , Constance A Owens 1 , Aashish C Gupta 2 , Suman Shrestha 1 , Tera S Jones 1 , Susan A Smith 1 , Kristy K Brock 1,2 , Laurence E Court 1 , Simona S Shaitelman 3 , Arnold C Paulino 3 , Chelsea C Pinnix 3 , James E Bates 4 , Rebecca M Howell 1 1 Department of Radiation Physics, The University of Texas MD Anderson Cancer Center, Houston, USA. 2 Department of Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, USA. 3 Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA. 4 Department of Radiation Oncology, Emory University, Atlanta, USA Purpose/Objective: Previous late effect studies on subsequent breast cancer in childhood/adolescent cancer survivors lack detailed breast volume dosimetry, as these survivors were predominantly treated during the pre-computed tomography (CT) era of radiotherapy. Although computational phantoms exist for dose reconstruction, they lack anatomically realistic breast models for pediatric anatomy, limiting the ability to incorporate dose-volume metrics for this population. We aimed to develop and validate a population-based pediatric breast model for reconstructing whole breast and quadrant doses for pre- and post-pubescent girls, introducing novel dose-volume metrics for late effect studies. Material/Methods: Breast contours were collected from whole-chest CT scans of 79 female Hodgkin lymphoma (HL) patients (12–21 years), with one selected as reference, 70 for training and eight for testing. A principal component analysis statistical shape modeling (PCA-SSM) method was performed on the training contours to capture population deformations. A mean breast model and two alternatives using the first five principal components were generated and integrated into an age-scalable computational phantom. Geometric evaluation assessed model compactness and generalizability through reconstructing test patient contours. Dosimetric accuracy was evaluated by computing breast doses for parallel-opposed HL 3D conformal mantle fields and comparing between test patients’ CT (ground truth) and corresponding age-matched phantoms with the PCA-SSM in Raystation. Dose-volume metrics were reported for whole breast and quadrants. Results: Reconstructing test patients’ contours yielded a mean (min-max) Dice similarity coefficient of 0.86 (0.85-0.88) and Hausdorff distance of 4.23mm (2.95-6.48mm), showing that the PCA-SSM accurately represents overall shape while capturing local variations. Average absolute differences for mean/max breast doses (normalized to prescription dose) between ground truth and mean model (alternative#1, alternative#2) were 2.63% (3.57%, 2.22%) and 2.80% (2.78%, 3.00%), respectively. Differences in whole breast dose-volume metrics (V5-V30) were typically within 5%, with some cases reaching 13%. For breast quadrants, mean dose differences were within 10% for all regions except the upper-inner and upper-outer quadrants, which reported differences of 16.12% (20.49%, 10.70%) and 10.06% (9.65%, 10.53%), respectively.