Investigation of Acuros XB radiotherapy dose calculation algorithm

Abstract

An important part of the radiotherapy process is the calculation of radiation dose deposition for individual patients. Accurate dose calculation provides a foundation for accurate dose prescription and reporting in radiation therapy. Several classes of radiation dose calculation algorithms exist. Some commonly used superposition-convolution algorithms are Collapsed Cone Convolution (CCC) (Philips Radiation Oncology Systems, Fitchburg, WI ) and Anisotropic Analytical Algorithm (AAA) (Varian Medical Systems, Palo Alto, CA, USA). Recently, Varian has introduced a new algorithm, Acuros XB, which solves the Linear Boltzmann Transport Equation (LBTE). This study assesses the accuracy of dose calculations performed with Acuros. In this study, a Monte Carlo model was created using the EGSnrc family of Monte Carlo code, in order to run Monte Carlo simulations. A Monte Carlo simulation involves simulating particle transport directly and is considered to be the gold standard in radiotherapy dose calculations. The Monte Carlo model is used to assess the accuracy of Acuros XB. In simple slab geometry, Acuros XB outperforms AAA when compared to Monte Carlo calculations and to measurements. Acuros XB and AAA show a dfference of -0.06 (+-) 0.13% and +0.6 (+-) 0.18% respectively, compared to measured doses distal to lung slabs of various thicknesses. Tests described in TecDoc 1583 [Vatnitsky and International Atomic Energy Agency, 2008] were used to validate Acuros and AAA calculations on a simple anthropomorphic phantom. Averaged over all TecDoc 1583 test cases and measurement points, AAA calculated dose 0.4 0.4% higher than measured, while Acuros calculated dose 0.2 (+-) 0.2% lower than measured. Calculations were made with Acuros, AAA and CCC for plans with small fields on Computed Tomography (CT) image sets of SBRT patients. The calculations were compared with Monte Carlo calculated dose. Acuros showed marginally better agreement with Monte Carlo than the other algorithms, for the patient plans and datasets used in this study. For a slice through the target volume of each plan, Acuros, AAA and CCC had a 98.0%, 97.8% and 96.7% gamma metric pass rate compared to Monte Carlo calculated dose, using the 2%, 2 mm gamma criteria.