ESTRO 37 Abstract book

S977

ESTRO 37

Results

MR/CT registration error and intrafraction motion are not included in this end-to-end test. EP-1814 New evaluation methodology of dose distributions for advanced calculation algorithms D. Jurado-Bruggeman 1 , A. Onsès-Segarra 1 , C. Muñoz- Montplet 1 1 Institut Català d'Oncologia, Medical Physics and Radiation Protection, Girona, Spain Purpose or Objective Advanced calculation algorithms, such as Monte Carlo or Boltzmann equation based, model radiation transport in media considering their composition and all interaction types. With respect to previous algorithms, accuracy improves but dose values can be significantly different hindering dose distribution evaluation and making hard to distinguish if these different doses are due to medium effects or to sub-optimal planning. The aim of this work is to propose an easy-to-implement methodology to aid dose distribution evaluation for these algorithms. For evaluation purposes, the calculated dose values are modified according to medium properties in each point. Material and Methods We used Acuros XB (AXB) v13.0 -Boltzmann equation-. Radiation transport is performed in medium to determine electron fluence in each point. The dose is obtained by using the physical properties -mass density and atomic composition- of the voxel: water or medium if the dose is reported in terms of water (Dw) or medium (Dm), respectively. We determined a curve to modify dose values depending on the medium. We used a virtual phantom with water density and a small adjustable heterogeneity in the middle (density 0-4.42 g/cm 3 ). Two parallel opposed fields of the same energy and MU provided uniform fluence. The dose conversion factor was obtained as the ratio between the doses for 1 g/cm 3 and for the considered density. We used 6 MV photons, AXB doses in terms of Dw (AXBw) and Dm (AXBm), and AAA algorithm for comparison. We tested the methodology in a virtual phantom with heterogeneities, and in 10 H&N clinical cases. We developed a MATLAB program that took the images and the dose matrix, and modified the dose value of each voxel by using its corresponding conversion factor. Then, we imported the modified dose matrix (AXBw* or AXBm*) into the Eclipse TPS. We programmed an Eclipse script to evaluate the following parameters for the PTV, with prescribed dose Dp and volume V: ●Homogeneity Index: (D2-D98)/Dp ●Homogeneity S-index=(∑[(Di-Dmean) 2 vi/V]) 1/2 , where vi is the ith volume element with dose ≥ Di ●D50% and D2%

Figure 1 shows how radiation transport characteristics remain while AXB doses in heterogeneities change. AXBm is more consistent but dose values are different.

Figure 2 shows AAA, AXB and AXB* differences in the evaluated parameters. AXBw distorts and hinders the quantitative evaluation of the dose distribution. AXBw* and AXBm solves most of the issues, but AXBm doses are systematically lower (D50%). Conclusion The proposed methodology serves as a bridge between advanced and classical algorithms for evaluation. For Dw, the improvements of these algorithms are kept while allows for: a more intuitive evaluation of dose distributions, the applicability of classical criteria - specially homogeneity and hot spots-, and differentiate if different dose values are due to medium effects or to sub-optimal planning. The use of Dm is more consistent, but further study is needed to determine prescription and OAR constraint values.