ESTRO 38 Abstract book

S477 ESTRO 38

institution’s TPS. 188 phantom irradiation plans (128 lung and 60 spine) were recomputed. The spine phantom simulates a highly modulated SBRT case; the lung phantom represents a low-to-none modulation moving target case. Results The results of the recalculation comparisons are shown in table 1. Overall, the DRS performed better than the TPS in about 52% of the spine cases. However, in the subset of phantoms that failed IROC criteria, the DRS was more accurate 93% of the time, indicating an error in the institution’s dose calculation system. In contrast, the lung phantom DRS recalculations were better only 31% of the time for all cases and 28% of the time in failing lung phantoms, indicating that dose calculation errors were not substantially present. Table 1. Performance of the dose recalculation system compared with institutions’ TPS. Figure 1 shows the differences between the accuracy of the DRS and TPS. These values represent the average percentage difference over all phantoms within each cohort. A positive difference % value indicates better DRS performance, meaning that the institution TPS is suboptimal. Correspondingly, a negative value indicates better TPS performance (which is the expected outcome if the TPS is well commissioned). In terms of magnitude of average dose accuracy (instead of frequency), the overall TPS performance surpassed that of the DRS in all cases except the failing spine phantoms which were more than 2% more accurate with the DRS as compared to the institution’s calculation. This reiterates the fact that for this phantom, a great majority of the error stems from the institutions’ dose calculation systems.

and 0.9%, respectively, showing largest sensitivity to low HU numbers for the Acuros and SciMoCa algorithms. The IQR for D 2 is nearly identical for all algorithms and <0.9% with median values <0.5%. Nearly no difference is seen between CBCT and CBCTlift for these near maximum doses. The median value of dose difference in D 1cc is <0.5% and the IQR is <1.6% for both oesophagus and spinal cord.

Conclusion Dose calculation on CBCT images calibrated by a stoichiometric metric and density override of HU values <950 results in clinically acceptable deviations for both target and OARs. A median deviation in target mean dose of <0.5% and IQR for D 98 and D 2 <1% was obtained for all algorithms. The lower quality in CBCT has a similar impact for all algorithms if HU override is applied. The CBCT calibration/modification method is readily implemented for automatic decision making for adaptive RT for all three dose computation algorithms PO-0900 The magnitude of dose calculation errors as a component of IROC phantom failures S. Edward 1 , M. Glenn 1 , P.A. Balter 1 , J.M. Pollard-Larkin 1 , C.B. Peterson 2 , R.M. Howell 1 , D. Followill 1 , S.F. Kry 1 1 The University of Texas MD Anderson Cancer Center, Radiation Physics, Houston, USA ; 2 The University of Texas MD Anderson Cancer Center, Biostatistics, Houston, USA Purpose or Objective From 2012 to present, 17% of IROC SBRT spine and 15% of IROC moving lung phantom irradiations have failed to meet established acceptability criteria. In this study, we looked at the contribution of dose calculation errors to We evaluated dose calculation errors by comparing the calculation accuracy of institutions’ treatment planning systems (TPS) versus IROC-Houston’s previously established [add reference] independent dose recalculation system (DRS). Each calculation was compared to the measured dose actually delivered to the phantom; cases where the recalculation was more accurate were interpreted as a deficiency in the these failing results. Material and Methods

Conclusion For spine phantoms, the institutions’ TPS performed better overall, but the DRS was remarkably better among the failing cohort. This indicates substantial room for improvement in many institutions’ TPS calculations (likely beam models). In contrast, lung phantom calculations indicated no direct dependence of failure on dose calculation. PO-0901 2D solid-state array detectors: a technique for in-vivo dose verification at varying effective area G. Biasi 1 , K. Utitsarn 1,2 , M. Petasecca 1 , M. Carolan 3 , V.L. Perevertaylo 4 , W.A. Tomé 5 , M.L.F. Lerch 1 , T. Kron 6,7 , A.B. Rosenfeld 1 1 University of Wollongong, Centre for Medical Radiation Physics, Wollongong, Australia ; 2 Lopburi Cancer Hospital, Department of Medical Services, Lopburi, Thailand ; 3 Wollongong Hospital, Illawarra Cancer Care Centre, Wollongong, Australia ; 4 SPA-BIT, n/a, Kiev,