Verification of a photon beam algorithm in a 3-D radiation therapy treatment planning system

We wanted to ensure that a three-dimensional photon beam dose calculation algorithm was correctly implemented in a commercially available radiation therapy treatment-planning system (TPS). The algorithm was verified by comparing interest-point dose values calculated by the TPS with dose values calculated manually. These comparisons were implemented in a variety of clinically relevant test cases. The initial reference test case consisted of a photon beam with a square field perpendicularly incident on a cubic water phantom. Interest points were located both along and off the central axis. All subsequent test cases were designed by modifying the reference case to allow for changes in field size and shape, beam angle of incidence, phantom heterogeneity, and other parameters along with the use of wedged and blocked fields. For 38 out of 41 interest points, dose values calculated by the TPS were within 1% of dose values calculated manually. For the three interest points in which TPS-calculated dose values deviated from manually calculated values by more than 1%, variances were attributed to inexact manual reproduction of TPS-dose calculation techniques and limitations of TPS input/output devices. No errors were detected in the computer code of the TPS. The series of tests described provides a viable method to ensure that a photon beam dose calculation algorithm is correctly implemented in the computer code of a TPS.