This study compared dose metrics between tangent breast plans calculated with the historical standard collapsed cone (CC) and the more accurate Monte Carlo (MC) algorithms. The intention was to correlate current plan quality metrics from the currently used CC algorithm with doses calculated using the more accurate MC algorithm.

Thirteen clinically treated patients, whose plans had been calculated using the CC algorithm, were identified. These plans were copied and recalculated using the MC algorithm. Various dose metrics were compared for targets and the time necessary to perform each calculation. Special consideration was given to V105%, as this is increasingly being used as a predictor of skin toxicity and plan quality. Finally, both the CC and MC plans for 4 of the patients were delivered onto a dose measurement phantom used to analyse quality assurance (QA) pass rates. These pass rates, using various evaluation criteria, were also compared.

Metrics such as the PTVeval D95% and V95% showed a variation of 6% or less between the CC and MC plans, while the PTVeval V100% showed variation up to 20%. The PTVeval V105% showed a relative increase of up to 593% after being recalculated with MC. The time necessary to perform calculations was 76% longer on average for CC plans than for those recalculated using MC. On average, the QA pass rates using 2%2mm and 3%3mm gamma criteria for CC plans were lower (19·2% and 5·5%, respectively) than those recalculated using MC.

Our study demonstrates MC-calculated PTVeval V105% values are significantly higher than those calculated using CC. PTVeval V105% is often used as a benchmark for acceptable plan quality and a predictor of acute toxicity. We have also shown that calculation times for MC are comparable to those for CC. Therefore, what is considered acceptable PTVeval V105% criteria should be redefined based on more accurate MC calculations.

To characterise small photon beams using the Monte Carlo dose calculation algorithm for small field ranges in a heterogeneous medium.

An in-house phantom constructed with three different mediums, foam, polymethyl methacrylate and delrin resembling the densities of lung, soft tissue and bone respectively, was used in this study. Photon beam energies of 6 and 15 MV and field sizes of 8×8, 16×16, 24×24, 32×32 and 40×40 mm using X-ray voxel Monte Carlo (XVMC) algorithm using different detectors were validated. The relative output factor was measured in three different mediums having six different tissue interfaces; at the depth of 0, 1, 2 and 3 cm. The planar dose verification was undertaken using gafchromic films and considered dose at the lung and bone medium interfaces. For all the measurements, 104×104 mm was taken as the reference field size. The relative output factor for all other field sizes was taken and compared with planning system calculated values.

From field size 16×16 mm and above, the relative output factors were analysed in bone and soft tissue medium having lung as first medium. The maximum deviations were observed as 1·8 and 1·3% for 6 MV and 2·5 and 1·1% for 15 MV photon beams for bone and soft tissue, respectively. For lung as measurement medium, the maximum deviation of 14·8 and 19·2% were observed and having bone as first medium with 8×8 mm for 6 and 15 MV photon beams, respectively. The fluence verification of dose spectrum for the lung–bone interface scenarios with smaller field sizes were found within 2% of deviation with treatment planning system (TPS).

The accuracy of dose calculations for small field sizes in XVMC-based treatment planning algorithm was studied in different inhomogeneous mediums. It was found that the results correlated with measurement data for field size 16×16 mm and above. Noticeable deviation was observed for the smallest field size of 8×8 mm with interfaces of significant change in density. The observed results demands further analysis of work with smaller field sizes.