ESTRO 2020 Abstract book
S100 ESTRO 2020
Studi di Firenze, Specialization School in Medical Physics, Firenze, Italy ; 3 Sant Joan de Reus Hospital, Department of Medical Physics, Tarragona, Spain ; 4 Hospital Clınic de Barcelona, Department of Radiation Oncology, Barcelona, Spain ; 5 Zhianawa Cancer Center, Zhianawa Cancer Center, Sulaymaniyha, Iraq Purpose or Objective Different clinical treatment sites can have inherent differences in the degree of plan complexity required to achieve a clinically acceptable plan. The delivery accuracy of a VMAT plan can be predicted by the score of plan modulation complexity. The aim of this work is to evaluate the influence of the detector resolution on the ability of various complexity metrics to predict delivery accuracy in VMAT treatments. Material and Methods 50 VMAT plans for three anatomical sites (17 Head&Neck, 19 Lung, 14 Breast) were considered. Pre‐treatment QA was performed with both the PTW Octavius1500 and Octavius729 detectors array inserted in the PTW Octavius4D phantom. Measurements were compared with TPS Elekta Monaco5.1 computed doses via 3D absolute dose γ‐analysis by using PTW VeriSoft 6.1. Global and Local 3D volumetric γ passing rate and γ mean evaluations with normalization at the 90% of the maximum dose and different criteria (3%/3mm, 3%/2mm, 2%/2mm) were performed. The complexity metrics (CM) evaluated from RT plans were: Modulation Complexity Score (MCS), Total Modulation Index (MITotal), the ratio of MLC side edge length and aperture area (EdgeMetric), the ratio between Leaf Travel and the Arc Length (LT/AL), meanMLCSpeed, MLCSpeedModulation, gantry speed modulation (GSmodulation), average leaf pair opening (mean MLC gap), plan averaged amounts of T&G effect (MeanTGi) and the ratio meanTGi/meanMLCgap. Spearman’s correlation coefficients between CM and γ evaluations were computed and statistical significance was assessed (Spearman’s rank test, p<0.01 and p<0.05). Results Spearman’s correlation coefficient showed high correlation between CM and γ valuations for some of the analysed CM (Tab. 1). When found, correlations were statistically significant on both γ passing rate and γ mean for both Local and Global Normalization. Octavius1500 allowed to highlight higher correlations with respect to Octavius729 for almost all the evaluated indices. The strongest correlations for Octavius1500 were found with MCS (>0.6 for all γ evaluations) and mean MLC gap (up to 0.799 for γ passing rate, Local Normalization). Octavius 729 results with local normalization showed lower correlation than Octavius1500 (fig.1). High correlations were found for LT/AL, meanMLCSpeed and MLCSpeedModulation indices with Octavius 1500 measurements for all γ evaluations. The same metrics showed weaker correlation, often not statistically significant, with Octavius 729.
Results Percentual dose differences (average ± σ) in ΔPTV Dmean were (3.1 ± 1.7)% for PIVD and (2.8 ± 1.7)% for EDIVD; in ΔCord DMax were (4.7 ± 3.3)% for PIVD and (10.3 ± 8.9)% for EDIVD and in Δ Parotid Dmean were (5.1% ± 3.1) for PIVD and (15.2 ± 10.1)% for EDIVD. In five cases Cord Dmax exceeded alarm level of 44 Gy with EDVD, but only three violations were find out with PIVD. In four cases, Parotid Dmean exceeded alarm level of 26 Gy, but only one violation was highlighted by PIVD.
Figure2: In this figure TPS and in vivo doses computed using plannig CT (PCT) e elastically deformed planing CT (EDCT) were compared in PTV, Cord and Parotid gland. Conclusion The use of EDCT for modeling patient anatomy improved the evaluation of dose differences in critical organ at risk. PCT produced results in agreement with EDCT for evaluation of dose difference in the PTV. PD-0183 Does the detector resolution influence correlations between complexity metrics and VMAT QA results? S. Russo 1 , G. Della Gala 2 , V. Hernandez 3 , M. Esposito 1 , S. Pini 1 , A. Ghirelli 1 , J. Saez 4 , H.O. Ghafour 5 , G. Zatelli 1 1 Azienda USL Toscana Centro, SC Fisica Sanitaria - Firenze, Bagno a Ripoli - Firenze, Italy ; 2 Università degli
Made with FlippingBook - Online magazine maker