Validation of Varian's AAA algorithm with focus on lung treatments

Abstract

Introduction. The objective of this study was to examine the accuracy of the Anisotropic Analytical Algorithm (AAA). A variety of different field configurations in homogeneous and in inhomogeneous media (lung geometry) was tested for the AAA algorithm. It was also tested against the present Pencil Beam Convolution (PBC) algorithm. Materials and methods. Two dimensional (2D) dose distributions were measured for a variety of different field configurations in solid water with a 2D array of ion chambers. The dose distributions of patient specific treatment plans in selected transversal slices were measured in a Thorax lung phantom with Gafchromic dosimetry films. A Farmer ion chamber was used to check point doses in the Thorax phantom. The 2D dose distributions were evaluated with a gamma criterion of 3% in dose and 3 mm distance to agreement (DTA) for the 2D array measurements and for the film measurements. Results. For AAA, all fields tested in homogeneous media fulfilled the criterion, except asymmetric fields with wedges and intensity modulated plans where deviations of 5 and 4%, respectively, were seen. Overall, the measured and calculated 2D dose distributions for AAA in the Thorax phantom showed good agreement – both for 6 and 15 MV photons. More than 80% of the points in the high dose regions met the gamma criterion, though it failed at low doses and at gradients. For the PBC algorithm only 30–70% of the points met the gamma criterion. Conclusion. The AAA algorithm has been shown to be superior to the PBC algorithm in heterogeneous media, especially for 15 MV. For most treatment plans the deviations in the lung and the mediastinum regions are below 3%. However, the algorithm may underestimate the dose to the spinal cord by up to 7%.