Abstract Book

S956

ESTRO 37

beam CT was used for set up correction of translational errors. In vivo three-dimensional dose distributions were computed by Dosimetry Check 4.1. The differences between planned and in vivo doses were evaluated using isocenter dose difference (ΔIso), Gamma Passing Rate (GPR) 3%/3mm (30% Dose threshold), and DVH differences (Δmin, Δmax, Δmean) in CTV. Tolerance for dose differences was set to 5% for mean CTV dose and 10% for maximum and minimum CTV doses. Statistical analysis was performed for evaluate correlations between GPR, DVH-differences and isocenter dose difference. Results 24% of fractions had DVH dose differences out of tolerance levels. In 46% of these fractions causes of discrepancy were identified. 38% of discrepancies were due to the incorrect position of immobilization devices, 31% was due to residual errors after correction of set-up by CBCT, and 31% was due to respiratory induced diaphragm motion. In the remaining fractions out of tolerance levels (54% of total), causes of discrepancy were not identified. Mean dose differences and standard deviation were (-2.6%± 3.5)% for Δmean, (-3.8%± 3.8)% for Δmin and (-3.6%± 3.8)% for Δmax in CTV; (-2.2%± 3.6)% for ΔIso. CTV Δmean, Δmin and Δmax resulted significantly correlated with ΔIso (p<0.01). GPR resulted uncorrelated with DVH and isocenter dose differences (p>0.25).

source treatment planning system matRad. Monte Carlo (MC) simulations using the Geant4/GATE 8.0 toolkit were performed to generate dose distributions maps in a 40x40x40 cm 3 water phantom within magnetic field regions up to 3T. Single monoenergetic parallel beams and a previously validated model for a proton research beam line were employed to generate calibration data for protons within the clinical energy range (62 - 252 MeV). In-depth longitudinal profiles, as well as parameters accounting for lateral deflection of the beam, beam broadening and transverse profiles anisotropy, were interpolated for different materials using a water-equivalent depth scaling. The performance of the algorithm was evaluated in homogeneous and heterogeneous phantoms filled with water, adipose and bone tissue. Dose distributions for central beams as well as volumetric targets using single-field pencil beam scanning proton plans were compared with MC simulations to assess the influence of magnetic fields in the analyzed treatment plans. Results For central beams, a close to perfect agreement was observed for calculations in water in magnetic fields of 0.5, 1.5T and 3.0T. IDD functions showed differences between the PBA and MC of less than 1% before the Bragg-Peak, and deviations of 2-8% in the distal energy falloff region. A good agreement was achieved using slab- like and lateral heterogeneous phantoms with maximal fluctuations in range of 0.7% and mean dose difference lower than 3%. Finally, treatment plans of comparable dosimetric quality to the implemented generic proton beam algorithm in matRad were obtained for box phantoms, see Figure 1.

Conclusion EPID in vivo dosimetry was able to identify treatment errors in abdominal and pelvic SBRT. GPR was not correlated with dose differences in DVH, otherwise in our analysis dose difference in isocenter was strongly correlated with all DVH dose differences analyzed. EP-1781 Towards treatment planning dose calculation for magnetic resonance guided proton therapy F. Padilla 1 , A. Resch 1 , D. Georg 1 , H. Fuchs 1 1 Medizinische Universität Wien, Department of Radiotherapy, Vienna, Austria Purpose or Objective After the successful clinical implementation of Magnetic Resonance Image (MRI) guided X-ray beam therapy, the possibility of MRI guidance for proton therapy (PT) is currently explored in a research setting. A pre-requisite for a successful implementation of MRI-PT is a proper treatment planning system. This work aims to implement and validate a dose calculation algorithm based on pencil beam kernels for proton beams within magnetic field regions. Material and Methods An in-house developed pencil beam algorithm based on a look-up table approach was implemented in the open-

Fig. 1 Spread out Bragg peak (SOPB) (a) and two dimensional dose distribution (b) for a cubic target in a box water phantom. The SOBP for a generic proton beam implemented in matRad is plotted for evaluation of the algorithm performance.