Abstract Book

ESTRO 37

S495

the feasibility of future MR-only workflows based on synthetic CT generation using tissue classification followed by bulk assignment as in the MRCAT (MR for Calculating ATtenuation) method. Material and Methods For CT scans of 6 head and neck patients with target volumes and OARs delineated, proton plans were calculated that were robust for 3 mm shifts and 3% range uncertainty. We evaluated the effect of different segmentation levels on the robustness of the proton dose distributions. In the first step, the original CT was segmented into varying number of bins to distinguish in increasing complexity water, air, fat and bone tissue. In the second step, Hounsfield unit (HU) values were bulk assigned, according to literature (see Table in Figure 1). In the more complex segmentations bone was segmented into cortical bone and marrow or the full HU range of bone was maintained (heterogeneous bone). Figure 1 shows the resulting segmented CTs. For one patient all segmentations were evaluated, for the other five patients the segmentations with only one HU for bone was abandoned based on the results of the first patient. The robust plans, planned on the full CT, were recalculated on the segmented CTs. The effects of the binning of the HU on the dose distribution were quantified by comparing the dose distributions of the segmented CT with that of the full CT by dose difference maps and gamma value maps.

Conclusion The results indicate that MR-only proton planning based on pseudo-CT is feasible given a stratification of bone HU into bone marrow and cortical bone is used and water, fat and air are segmented. Further evaluation on a larger cohort of patients is needed to strengthen our findings. PO-0921 Local control rates in stereotactic body radiotherapy of lung metastases related to dose algorithm D. Zucca Aparicio 1 , O. Hernando Requejo 2 , M.A. De la Casa de Julián 1 , C. Rubio Rodríguez 3 , P. Fernández Letón 1 1 Hospital Universitario HM Sanchinarro, Servicio de Radiofísica y Protección Radiológica, Madrid, Spain 2 Hospital Universitario HM Puerta del Sur, Servicio de Oncología Radioterápica, Móstoles, Spain 3 Hospital Universitario HM Sanchinarro, Servicio de Oncología Radioterápica, Madrid, Spain Purpose or Objective To report the correlation of the biologic effective dose (BED) for SBRT of lung metastases with the tumor control probability (TCP) to estimate the BED threshold above which successful local control is found. To evaluate the dose differences regarding the algorithm, target volume and tissue density. Material and Methods Forty-seven metastases were calculated with Pencil Beam (PB) and Monte Carlo (MC) from iPlan to obtain the differences between both algorithms for the dose to the 99% (D 99% ) and 50% (D 50% ) of PTV. Adapted risk-dose schemes have been used according to the location of the lesion or nearness to critical structures. The dose was prescribed to the 95% of PTV. The plans were calculated on breath-hold CT scans in exhale, with the same beam arrangement and MU for both algorithms. The PB algorithm was used for dose calculation until the clinical implementation of MC in 2011. The CTV was contoured in the lung/mediastinum windows, with a 5-mm margin. The treatment delivery with respiratory gating in a Novalis was based on internal markers to correlate tumour motion with breathing. The BED to 99% (BED 99% ) and 50% (BED 50% ) of PT V was estimated from MC results. Local control was evaluated after 24 m of median follow-up [3-52 m]. TCP was obtained by adjusting the observed local control data to a sigmoid dose-response curve. Results The linear regression analysis regarding the dose deviations between algorithms, the inverse of the volume and the electron density of PTV is summarized in Table 1. The greatest deviations (40.0% in D 99% and 38.4% in D 50% )

Results A single segmentation for bone resulted in dose differences above 10% and gamma values larger than 2. However, a WFAMCB segmentation with 5 HU-values in total and with 2 HU-values for bone (with a differentiation between bone marrow and cortical bone), shows dose differences smaller than 5% and gamma values hardly exceeding 1. In fact, we found gamma pass rates larger than 96% with a gamma criterion of 1%/1mm when comparing the dose distribution recalculated on the segmented CT with the dose distribution on the full CT (Table 1). Dose differences in the target regions were 0.1% or less, dose differences in high-gradient regions in Organs at Risk (OAR) were less than 4%.