Quantification of dose differences between two versions of Acuros XB algorithm compared to Monte Carlo simulations — the effect on clinical patient treatment planning

Abstract

A commercialized implementation of linear Boltzmann transport equation solver, the Acuros XB algorithm (AXB), represents a class of most advanced type ‘c’ photon radiotherapy dose calculation algorithms. The purpose of the study was to quantify the effects of the modifications implemented in the more recent version 11 of the AXB (AXB11) compared to the first commercial implementation, version 10 of the AXB (AXB10), in various anatomical regions in clinical treatment planning. Both versions of the AXB were part of Varian's Eclipse clinical treatment planning system and treatment plans for 10 patients were created using intensity‐modulated radiotherapy (IMRT) and volumetric‐modulated arc radiotherapy (VMAT). The plans were first created with the AXB10 and then recalculated with the AXB11 and full Monte Carlo (MC) simulations. Considering the full MC simulations as reference, a DVH analysis for gross tumor and planning target volumes (GTV and PTV) and organs at risk was performed, and also 3D gamma agreement index (GAI) values within a 15% isodose region and for the PTV were determined. Although differences up to 12% in DVH analysis were seen between the MC simulations and the AXB, based on the results of this study no general conclusion can be drawn that the modifications made in the AXB11 compared to the AXB10 would imply that the dose calculation accuracy of the AXB10 would be inferior to the AXB11 in the clinical patient treatment planning. The only clear improvement with the AXB11 over the AXB10 is the dose calculation accuracy in air cavities. In general, no large deviations are present in the DVH analysis results between the two versions of the algorithm, and the results of 3D gamma analysis do not favor one or the other. Thus it may be concluded that the results of the comprehensive studies assessing the accuracy of the AXB10 may be extended to the AXB11.PACS numbers: 87.55.‐x, 87.55.D‐, 87.55.K‐, 87.55.kd, 87.55.Qr

Validation of the Acuros XB dose calculation algorithm versus Monte Carlo for clinical treatment plans

“…Most of these studies were based on former versions of Acuros (v10 or v11). For each new version, minor improvements are made to the algorithm . The findings of the former studies differ considerably, showing large discrepancies between Acuros and MC calculations for a number of cases.…”

“…A comparison to MC for 10 patients treated at different anatomical regions and with different treatment techniques found GAI below 90% for the PTV in four of the patients. The low GAI values were found for brain, breast, lung, and vaginal cancer patients . Most of these studies were based on former versions of Acuros (v10 or v11).…”

“…The low GAI values were found for brain, breast, lung, and vaginal cancer patients. 15 Most of these studies were based on former versions of Acuros (v10 or v11). For each new version, minor improvements are made to the algorithm.…”

“…Dosimetric one‐dimensional (1D) and two‐dimensional (2D) measurements and MC calculations have been used for the validation. MC‐based benchmarking has been performed using CT datasets of 4 lung cancer patients and 10 patients treated at a variety of anatomical sites . In both studies, the gamma agreement index (GAI) with 2% dose difference and 2 mm distance‐to‐agreement (DTA) were reported, and the GAI ranged from 82% in a patient with lung cancer to 98% in a patient with pancreas cancer.…”

“…MC-based benchmarking has been performed using CT datasets of 4 lung cancer patients 14 and 10 patients treated at a variety of anatomical sites. 15 In both studies, the gamma agreement index (GAI) with 2% dose difference and 2 mm distance-to-agreement (DTA) were reported, and the GAI ranged from 82% in a patient with lung cancer to 98% in a patient with pancreas cancer. Given these studies, it can be concluded that in previous version, Acuros has been on a par with other advanced algorithms (convolution/superposition, collapsed cone), but it has not exceeded their accuracy to reach the level of a well-implemented MC algorithm.…”