ESTRO 2020 Abstract book

S278 ESTRO 2020

magnetic field transverse to the radiation beam compounds bulk ED-associated dosimetric discrepancies caused by the electron return effect in heterogeneous tissues. An estimation of the errors associated with a multiple-ROI bulk ED treatment planning approach was performed for MRL partial breast radiotherapy in order to assess clinical suitability. Material and Methods Four partial breast cancer (2 left-sided, 2 right-sided) patient CTs and structures were imported to Monaco 5.40.01 (Elekta). MRL IMRT plans (40.05 Gy / 15 fractions) were created using the CT Hounsfield-Unit to ED lookup table (LUT) for dose calculation. Doses were recalculated with patient-specific mean relative EDs assigned to ROIs of bones, CTV, heart, lungs, external. The range of relative ED values (relative to water) for each ROI across patients was investigated. Dose distributions for the CT LUT and bulk ED calculations were compared using dose differences, global gamma index and DVH metrics: PTV D95%; heart mean dose and ipsilateral lung D15%. Results Relative EDs for ROIs, averaged across all patients were: bones 1.159 (1.141 – 1.194), CTV 0.926 (0.910 – 0.973), heart 1.031 (1.026 – 1.038), lungs 0.292 (0.214 – 0.379), external 0.947 (0.934 – 0.961). Mean and max absolute percentage differences in DVH metrics resulting from the use of bulk ED were found to be negligible: PTV D95% (mean 0.3%; max 0.7%), heart mean dose (mean 0.2%; max 0.8%), ipsilateral lung D15% (mean 0.9%; max 1.6%). The dose difference distributions between CT LUT and bulk ED calculations showed variation predominantly within -5% to +5% of 40.05 Gy. The bulk ED right-sided plans exhibited underestimation of the dose by up to 10% in lung proximal to the treated region, while elsewhere in the ipsilateral lung, dose was moderately over-estimated (approx. 2%). The opposite pattern was observed for the left-sided cases. Gamma evaluation of the CT LUT and bulk ED calculated doses across all patients gave average pass rates (γ index < 1) for voxels receiving > 20% of 40.05 Gy in the CT LUT dose distribution of 0.954 and 0.862 for criteria 3% / 3 mm and 2% / 2 mm respectively (see Table 1). Failing gamma voxels were highly concentrated in the lung and chest wall proximal to the target (Figure 1).

Results The tungsten sphere was found to have the highest absorption of the investigated spheres (Figure 1) and therefore was found to be best suited for irradiation purposes. The magnetometer measurements revealed a volume susceptibility of χ = (9.43 ± 0.45) ∙ 10 -6 (SI units) for tungsten. The maximum distortions within the distortion maps were calculated and validated the trend of decreasing values with increasing salt concentration. In general, a higher extent of distortions was found at 1.5 T as compared to 0.35 T. At 0.35 T the addition of copper sulfate (100 g/l) reduced the artefacts induced by the tungsten sphere by 30 %. At 1.5 T a reduction by 61 % was achieved using the highest soluble amount of copper sulfate (200 g/l). In all measurements, the maximum distortions did not exceed 1.7 mm and were located in close vicinity of the sphere.

Conclusion In this work, a method to reduce susceptibility-induced artefacts in MRI at interfaces between a metal sphere and water was developed. By adding salt to the water, a maximum reduction by 61 % was achieved. This is especially important for a quality assurance phantom used for isocenter validation in MRgRT but can be used in other phantoms as well. OC-0465 Evaluation of the use of electron density overrides in MR-Linac breast treatment planning J. Mohajer 1 , A. Mitchell 1 , A. Dunlop 1 , S. Nill 1 , U. Oelfke 1 1 The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust, Joint Department of Physics, London, United Kingdom Purpose or Objective Bulk electron density (ED) assignment to regions of interest (ROIs) represents one solution to MR-based dose calculation. For tissues with high ED heterogeneity, e.g. lung, bulk ED assignment may introduce significant dosimetric errors. Treatment delivery on the Unity MR- Linac (MRL, Elekta AB, Stockholm) in a 1.5 T static