The accuracy of dose calculations by anisotropic analytical algorithms for stereotactic radiotherapy in nasopharyngeal carcinoma.

Nasopharyngeal tumors are commonly treated with intensity-modulated radiotherapy techniques. For photon dose calculations, problems related to loss of lateral electronic equilibrium exist when small fields are used. The anisotropic analytical algorithm (AAA) implemented in Varian Eclipse was developed to replace the pencil beam convolution (PBC) algorithm for more accurate dose prediction in an inhomogeneous medium. The purpose of this study was to investigate the accuracy of the AAA for predicting interface doses for intensity-modulated stereotactic radiotherapy boost of nasopharyngeal tumors. The central axis depth dose data and dose profiles of phantoms with rectangular air cavities for small fields were measured using a 6 MV beam. In addition, the air-tissue interface doses from six different intensity-modulated stereotactic radiotherapy plans were measured in an anthropomorphic phantom. The nasopharyngeal region of the phantom was especially modified to simulate the air cavities of a typical patient. The measured data were compared to the data calculated by both the AAA and the PBC algorithm. When using single small fields in rectangular air cavity phantoms, both AAA and PBC overestimated the central axis dose at and beyond the first few millimeters of the air-water interface. Although the AAA performs better than the PBC algorithm, its calculated interface dose could still be more than three times that of the measured dose when a 2 × 2 cm(2) field was used. Testing of the algorithms using the anthropomorphic phantom showed that the maximum overestimation by the PBC algorithm was 20.7%, while that by the AAA was 8.3%. When multiple fields were used in a patient geometry, the dose prediction errors of the AAA would be substantially reduced compared with those from a single field. However, overestimation of more than 3% could still be found at some points at the air-tissue interface.