S2043
Physics - Image acquisition and processing
ESTRO 2026
Digital Poster Highlight 474 Unlocking simulation-free oART using diagnostic CT: impact of kV and calibration phantom Miguel Martínez Albaladejo 1 , Ana Corbalán Mirete 1 , Aitor Ortega González 1 , Johnathan Suárez Arteaga 1 , David Ramos Amores 1 , Vicente Puchades Puchades 1 , Irene Císcar García 2 , Antonio Javier García Sánchez 3 , Alfredo Serna Berná 1 1 Medical Physics and Radiation Protection, Complejo Hospitalario Universitario de Cartagena, Cartagena, Spain. 2 Oncology Radiation Department, Complejo Hospitalario Universitario de Cartagena, Cartagena, Spain. 3 Department of Information and Communication Technologies, Universidad Politécnica de Cartagena, Cartagena, Spain Purpose/Objective: This study aims to evaluate the dosimetric impact of CT peak-voltage (kVp) and phantom selection on HU- to-density conversion for IMRT/VMAT planning, using Acuros XB (AXB) and Anisotropic Analytical Algorithm (AAA). To our knowledge, this is the first work to assess the combined impact of diagnostic CT (dCT) kVp variation and phantom choice for simulation-free online adaptive radiotherapy (oART). Material/Methods: Both HU-to-relative electron density (RED) and HU-to- mass density (MD) calibration curves were generated using two phantoms—CIRS 062M (tissue-equivalent) and CatPhan 504 (plastic-based)—scanned at 80–140 kVp on a Siemens go.Sim CT. 20 palliative 6FFF-Ethos- VMAT/IMRT plans were recalculated in Eclipse v16 using these calibration curves with AXB (MD- based)/AAA (RED-based). Two dosimetric comparisons were conducted: (1) evaluating the kVp effect using kVp-specific calibrations (kVp=120 as departmental reference) for each phantom, and (2) contrasting phantom types across kVp levels (CIRS as reference). Evaluation included DVH metrics (PTV- V95%, V105%; Dmedian; OAR-Dmean; body-Dmax; conformity index-CI). Additionally, 𝛾 -analysis was performed by comparing passing rates and mean 𝛾 values between dose distributions calculated under reference conditions and those varying kVp and phantom (2%/2mm, 1%/1mm). Statistical significance was assessed using Wilcoxon signed-rank and two-way ANOVA tests ( α =0.05). Results: kVp-induced HU variations were minimal in CatPhan but pronounced in CIRS high-density inserts, where they were accurately modelled by second-order polynomial fits (R2 > 0.999; Figure 1). Despite this kVp- related HU variability, the dosimetric impact was negligible (< 1 %) across all phantoms and algorithms (Figure 2), with AXB demonstrating slightly greater robustness than AAA. Median γ -passing rates for
Conclusion: The SynthRAD2025 Grand Challenge provided a large- scale, multi-institutional benchmark for sCT generation in complex anatomical sites. It demonstrated that modern deep learning models can produce sCTs of high image and dosimetric accuracy for both MRI and CBCT inputs, facilitating MRI-only and adaptive radiotherapy protocols. References: 1 Spadea MF & Maspero M, Zaffino P, Seco J. Deep learning-based synthetic-CT generation in radiotherapy and PET: A review. Medical Physics. 2021, 48(11), 6537–6566. https://doi.org/10.1002/mp.151502 Thummerer A, van der Bijl E, ..., Maspero M. SynthRAD2025 Grand Challenge dataset: Generating synthetic CTs for radiotherapy from head to abdomen. Med Phys. 2025 ;52(7):e17981. https://doi.org/10.1002/mp.179813 Wieser HP, Cisternas E, Wahl N, Ulrich S, Stadler A, Mescher H, ... & Bangert M. (2017). Development of the open - source dose calculation and optimization toolkit matRad. Med Phys, 44(6), 2556- 2568. https://doi.org/10.1002/mp.12251 Keywords: synthetic CT, deep learning, MRI and CBCT
Made with FlippingBook - Share PDF online