ESTRO 38 Abstract book

S964 ESTRO 38

2mm, we cannot assure that one matrix results are superior to the other ones.

1 Consorci Sanitari de Terrassa CST, Radiotherapy, Terrassa, Spain

Purpose or Objective With the increasing use of Stereotactic Body Radiation Therapy (SBRT) combined with VMAT technique the complexity of treatments has increased significantly. Now the dose distributions are less homogenous than with 3DCRT and the dose gradients are higher. We need to calculate and measure these dose distributions as accurately as possible. The first objective of this work is to compare the dose distributions measured with two commercial array detectors against the calculated dose in Monaco TPS. The second objective is to evaluate the influence of the grid spacing used in the TPS calculation for the SRS1000 detector. In both we analyze lung and spinal SBRT cases separately. Material and Methods The SBRT plans were calculated in Monaco 5.10 TPS and did the QA with 1% statistical uncertainty and 2 and 1mm grid spacing. All the plans were delivered with an Elekta Synergy with Agility ML linac. We verified both 29 lung and 6 spinal SBRT cases. The QA measurements were done with the PTW Octavius4D 1500 (1405 Ionization chambers (ICs) of 0.06 cm3 arranged in a 27x27cm matrix with a detector spacing of 7.1 mm) and PTW Octavius4D 1000SRS (977 liquid-filled ICs of 0.003 cm3 arranged in a 10x10cm matrix with a detector spacing of 2.5 mm in the center area). We used the 3D gamma index γ(%dose,DTA) to compare the measured and calculated dose distributions. Due to the special characteristics of SBRT treatments, we used γ(2%,2mm) and γ(3%,1mm) for the evaluations in the grid spacing case. In the 1500vs1000 comparison we evaluated just the γ(2%,2mm). Results Results are shown in Table 1. For the statistical comparison in both Lung and Spinal patients we first evaluated if the data was normally distributed. Lung data resulted not normally distributed so we did a Mann- Whitney-Wilcoxon test. Spinal data were normally distributed so we did a paired t test with a 95% confidence interval. For the 1000SRS grid comparison (2 vs 1mm) we found in both lung and spinal patients that the differences in the comparisons were statistically significant. For the 1500 vs 1000SRS gamma comparison, we had the same conclusion although the significance was weak.

EP-1783 Implementation of a fast method for routine linac-QA for VMAT with EPID dosimetry P. Cambraia Lopes 1 , J. Van Egmond 1 , M. De Goede 1 , J. Van Santvoort 1 1 Haaglanden Medisch Centrum- Antoniushove, Medical Physics, Leidschendam, The Netherlands Purpose or Objective VMAT plans are rather complex and require continuous and synchronized change of many variables in time. It would therefore be desirable to regularly assess the quality of dynamic delivery, independent of any particular (patient) plan. We have developed a fast and simple method intended to detect suboptimal VMAT delivery by the linac, VMAT test plans were developed that contained relatively extreme dose rates (DR, range: ~65-500 MU/min), gantry speeds (GS, range: ~1-5 deg/sec), and MLC-leaf speeds (LS, up to 3.5 cm/sec), which can be found in clinical practice. The following linac states were constructed for the Pinnacle treatment planning system (TPS), by specifying for each VMAT segment the corresponding MU’s and leaf positions: High-DR/High-GS (HH, 1.67 MU/deg), High-DR/Low-GS (HL, 10 MU/deg), and Low-DR/High-GS (LH, 0.22 MU/deg). Plans contained sharp transitions between these states, which also allowed to test for inertia effects (acceleration/deceleration) of the gantry, DR control, and for proper synchronization. Beam 1 contains linac states HH/HL, together with MLC movement, while beam 2 contains linac states HL/LH with static MLC (figure 1). The gantry travel between linac states was varied and kept as a multiple of 2 degrees. These plans were delivered on a cylindrical phantom while EPID frames were recorded. The delivered dose was then reconstructed with the iViewDose software from Elekta and the reconstructed dose was compared in 3D to the TPS dose. In addition, the test plans were benchmarked by independent dose measurements using a dedicated phantom for pretreatment patient-QA (ArcCHECK phantom), and by monitoring the linac parameters DR, GS, and LS, during delivery. Results It is necessary to explicitly define control points (CP) every 2 degrees, to get sufficiently detailed dose calculations from the TPS. By doing so, the agreement between the benchmark dose measurement and the TPS dose was found to be 78% for beam 1, and 95% for beam 2 (% gamma pass rate with criteria 3%/3mm). The agreement between iViewDose and the TPS was 97% for beam 1 (mean gamma using EPID dosimetry. Material and Methods

Conclusion Comparing the 1000SRS measurements with 2 and 1mm calculation grids, we can conclude that for spinal cases 1mm grid represents better the dose delivered than 2mm grid. However, in lung cases that does not happen. We think that this happens because both dose distributions are in general very different. While in spinal SBRT we usually have large dose concavities with the corresponding dose fall-off, in lung SBRT we usually have approximately- rounded targets, were we hypothesize that spatial resolution is not that extremely crucial in calculations (Figure 1). In spite of that, both lung and spinal comparisons show statistically different results in both γ(2%,2mm) and γ(3%,1mm). In the 1500 vs 1000SRS comparison, assuming a grid of