Accuracy of the Small Field Dosimetry Using the Acuros XB Dose Calculation Algorithm within and beyond Heterogeneous Media for 6 MV Photon Beams

Purpose: The dosimetric accuracy of the recently released Acuros XB advanced dose calculation algorithm (Varian Medical Systems, Palo Alto, CA) is investigated for single radiation fields incident on homogeneous and heterogeneous geometries, as well as for two arc (VMAT) cases and compared against the analytical anisotropic algorithm (AAA), the collapsed cone convolution superposition algorithm (CCCS) and Monte Carlo (MC) calculations for the same geometries. Methods and Materials: Small open fields ranging from 1 × 1 cm to 5 × 5 cm were used for part of this study. The fields were incident on phantoms containing lung, air, and bone inhomogeneities. The dosimetric accuracy of Acuros XB, AAA and CCCS in the presence of the inhomogeneities was compared against BEAMnrc/DOSXYZnrc calculations that were considered as the benchmark. Furthermore, two clinical cases of arc deliveries were used to test the accuracy of the dose calculation algorithms against MC. Results: Open field tests in a homogeneous phantom showed good agreement between all dose calculation algorithms and MC. The dose agreement was +/−1.5% for all field sizes and energies. Dose calculation in heterogenous phantoms showed that the agreement between Acuros XB and CCCS was within 2% in the case of lung and bone. AAA calculations showed deviation of approximately 5%. In the case of the air heterogeneity, the differences were larger for all calculations algorithms. The calculation in the patient CT for a lung and bone (paraspinal targets) showed that all dose calculation algorithms predicted the dose in the middle of the target accurately; however, small differences (2% 5%) were observed at the low dose region. Overall, when compared to MC, the Acuros XB and CCCS had better agreement than AAA. Conclusions: The Acuros XB calculation algorithm in the newest version of the Eclipse treatment planning system is an improvement over the existing AAA algorithm. The results are comparable to CCCS and MC calculations especially for both stylized and clinical cases. Dose discrepancies were observed for extreme cases in the presence of air inhomogeneities.