ESTRO 36 Abstract Book

S475 ESTRO 36 _______________________________________________________________________________________________

Purpose or Objective To analyze the effects of considering the real distribution of random errors (s) within our patient population in the outcome of setup correction protocols. Results are compared to those predicted by Van Herk`s margin formula (VHMF) considering constant random errors. Material and Methods Displacement data from 31 prostate and 31 head and neck (HN) treatments were employed in this study, based on 640 and 540 CBCT images respectively. Values of σ at each direction were calculated by obtaining the standard deviation of the corrections during the treatment of each patient. The proposed distribution for the modelling of heterogeneous σ 2 is an IG distribution (eq. 1). This kind of distribution has been demonstrated to be suitable for modelling random errors (Herschtal et al, Phys Med Biol 2012:57:2743-2755). Parameters a and b of the IG distribution can be obtained from the mean value and standard deviation of the measured σ 2 distribution. Treatment margins proposed by VHMF for a No Action Level using the first 5 fractions for setup correction (NAL 5) protocol were obtained by considering a constant σ for all the patients. Given the margins proposed, the patient coverage for the real σ distribution was obtained by weighting the dose coverages for each combination of σ values at each direction with the probability that a patient has those values of σ, based on the fitted IG distributions. Results Results are shown in Table 1. It can be seen that, if heterogeneities in random error distribution are taken into account, the coverage probability yields values smaller than those predicted by VHMF when homogenous σ is considered. After this results were obtained, calculations for different sets of margin were done. It was found that in the HN case, margins had to be increased 1 mm at each direction to obtain coverages of a 92 %, while in the prostate case, margins had to be increased 1.4 mm in all directions in order to achieve a coverage of the 90%. These results suggest that the effects of heterogeneous random errors depend on the characteristics of the random error distribution of the patient population.

Purpose or Objective Esophagus and the organs at risk (OAR) nearby move with respiration. The purpose of this study was to determine if respiratory gating or deep inspiration breath hold (DIBH) facilitate dose reduction to OAR. Material and Methods CT image sets from ten patients were analysed. Esophagus and OAR were delineated on end expiration (EE) and end inspiration (EI) phases of the 4DCT and on DIBH CT. 5 cm long mock GTVs were delineated in the proximal (P), medial (M) and distal (D) part of the esophagus. CTVs were defined by expanding the GTVs according to our clinical practice. CTV to PTV margin was 7 mm. Relative position of OARs and target were evaluated with cumulative distance volume histograms (DiVHs) [Wu et al. Med Phys 2009], calculated for the part of the OAR located in the beam path. The most and least optimal phase for treatment was selected by comparing the percent of the OAR volume located within the distance intervals A (below 2.5 cm), B (2.5-5.0 cm) and C (5.0-7.5 cm) from the PTV. The organ sparing achieved or lost, by changing treatment from FB to a specific breathing phase, was estimated by assuming that FB can be simulated with 50% EE and 50% EI. Results Esophagus elongation during 4DCT was median 11mm (range 2-20mm) and from EE to DIBH 23mm (10-42mm). Lung volume increased 13.3% (6.9-24.9%) from EE to EI and 63.5% (34.1-120.8%) from EE to DIBH. Absolute volume of lung in the beam path either increased or remained largely constant upon inspiration in all patients. In seven P, four M and four D targets, the absolute volume of lung located within 5 cm of the PTV increased; however, increase in the total lung volume still resulted in either a reduction or a largely unchanged percent lung volume located within 5 cm of the PTV. Results extracted from DiVHs are presented in Table 1.

Conclusion The effect of heterogeneous random errors should be taken into account when applying treatment margins, its effects depend on the characteristics of the patient population and should be analyzed for each treatment location at each institution. PO-0872 Respiration motion management strategy for sparing of risk organs in esophagus cancer radiotherapy S.B.N. Biancardo 1 , J.C. Costa 1 , K.F. Hofland 1,2 , T.S. Johansen 1 , M. Josipovic 1 1 Rigshospitalet, Department of Oncology- Section of Radiotherapy, Copenhagen, Denmark 2 Zealand University Hosptial, Department of Oncology- Section of Radiotherapy, Naestved, Denmark

DIBH was the optimal treatment phase for all P and M targets and 8/10 D targets. For all targets EE was the least optimal phase. Heart displacement was ≤12mm on 4DCT and ≤26mm from EE to DIBH. Relative heart volume DiVH’s are shown in Figure 1 for 3 patients. The same respiration phase is clearly not optimal for all patients, neither for M nor for D targets. EE was most optimal for heart sparing in two M and three D, EI in four M and three D and DIBH in four M and four D targets. EE was least optimal in four M and two D, EI in two M and three D and DIBH in four M and five D targets.