S2349
Physics - Quality assurance and auditing
ESTRO 2026
were analysed, including 12 lung, 5 head and neck, 9 brain, 7 bone SBRT, 10 breast, 5 rectum, and 5 prostate cases. For each site, relevant OARs were specifically evaluated: breast (heart, ipsilateral lung, contralateral breast), bone SBRT (spinal cord, cauda equina), brain (spinal cord, brainstem, brain), head and neck (spinal cord, parotids), lung (heart, spinal cord, ipsilateral lung), rectum (bowel, bladder), and prostate (rectum, bladder). Global gamma evaluation (3%/3 mm, 10% dose threshold, ≥ 90% passing points) and median dose deviation (%) were analysed by anatomical structure to assess clinical applicability and performance. Results: Simple open-field measurements demonstrated high agreement between measured and planned doses, confirming the fundamental accuracy of the Delta ⁴ system (Table 1). When applying the DVH Anatomy model to clinical VMAT data, larger deviations were observed, mainly for target structures. Targets showed increased median dose discrepancies, more pronounced for X6MV energy, suggesting a possible limitation in the current algorithm calibration rather than an effect of plan modulation. In contrast, OARs presented more consistent results with higher γ (3%,3 mm) passing rates and smaller dose deviations, indicating better modeling performance in low- gradient regions (Figure 1).
each MP with ADDI was 98.9% (96.7–99.9). In non- inferiority testing, ADDI’s mean Δκ for the final verdict was 0.0006 (95% CI: − 0.0073–0.0083, p << 0.05) (Figure 2), indicating that ADDI’s performance was non- inferior to human raters. Conclusion: Substantial inter- and intra-observer variability was observed among MPs during plan evaluation, highlighting subjectivity in current methods. ADDI’s final verdict agreement with the ground truth was within the range of inter-rater variability. Non- inferiority testing showed that ADDI’s substitution for an MP did not significantly alter group consensus. Therefore, ADDI can reliably emulate expert-level decision-making and support consistent, automated plan evaluation in routine clinical workflows. References: [1] Olanrewaju A et al.,Pract Radiat Oncol. 2021;11:177–184. doi:10.1016/j.prro.2020.12.003.[2] Gueiderikh A et al.,Front Oncol. 2023;13:1199043. doi:10.3389/fonc.2023.1199043.[3] Hansen CR et al.,J Med Imaging Radiat Oncol. 2022. doi:10.1111/1754- 9485.13374.[4] Maladenige HC et al.,Phys Med Biol. 2025. doi:10.1088/1361-6560/ae11f4.[5] Virén T et al.,Radiat Oncol. 2015;10:79. doi:10.1186/s13014-015- 0392-x. Keywords: advanced dose distribution index, ADDI, quality Assessing the Clinical Value of Delta⁴ ScandiDos DVH Anatomy for routine Patient-Specific Quality Assurance Sara Poeta, Youssef Ez-zyouy, Younes Jourani Department of Medical Physics, Université Libre de Bruxelles (ULB), Hôpital Universitaire de Bruxelles (H.U.B), Institut Jules Bordet, Brussels, Belgium Purpose/Objective: Traditional patient-specific QA using the gamma index lacks anatomical and clinical relevance1. The Delta ⁴ ScandiDos DVH Anatomy module enables dose evaluation by structure, linking deviations to targets and OARs for more meaningful analysis. The aim of this work is to assess its clinical value and practicality for routine patient-specific QA implementation. Material/Methods: To optimise calculation accuracy in Delta ⁴ ScandiDos DVH Anatomy, detailed beam data from our linear accelerators—including dose profiles, output factors in air and water, and related parameters—were provided Digital Poster 1529 to ScandiDos for algorithm refinement. Initial validation was performed using simple anterior– posterior and lateral fields for each energy, to verify system response. Subsequently, clinical VMAT plans
Made with FlippingBook - Share PDF online