Comparison of the ESTRO formalism for monitor unit calculation with a Clarkson based algorithm of a treatment planning system and a traditional "full-scatter" methodology.

The ESTRO formalism for monitor unit (MU) calculations was evaluated and implemented to replace a previous methodology based on dosimetric data measured in a full-scatter phantom. This traditional method relies on data normalised at the depth of dose maximum (Zm), as well as on the utilisation of the BJR 25 table for the conversion of rectangular fields into equivalent square fields. The treatment planning system (TPS) was subsequently updated to reflect the new beam data normalised at a depth ZR of 10 cm. Comparisons were then carried out between the ESTRO formalism, the Clarkson-based dose calculation algorithm on the TPS (with beam data normalised at Zm and ZR), and the traditional "full-scatter" methodology. All methodologies, except for the "full-scatter" methodology, separated head-scatter from phantom-scatter effects and none of the methodologies; except for the ESTRO formalism, utilised wedge depth dose information for calculations. The accuracy of MU calculations was verified against measurements in a homogeneous phantom for square and rectangular open and wedged fields, as well as blocked open and wedged fields, at 5, 10, and 20 cm depths, under fixed SSD and isocentric geometries for 6 and 10 MV. Overall, the ESTRO Formalism showed the most accurate performance, with the root mean square (RMS) error with respect to measurements remaining below 1% even for the most complex beam set-ups investigated. The RMS error for the TPS deteriorated with the introduction of a wedge, with a worse RMS error for the beam data normalised at Zm (4% at 6 MV and 1.6% at 10 MV) than at ZR (1.-9% at 6 MV and 1.1% at 10 MV). The further addition of blocking had only a marginal impact on the accuracy of this methodology. The "full-scatter" methodology showed a loss in accuracy for calculations involving either wedges or blocking, and performed worst for blocked wedged fields (RMS errors of 7.1% at 6 MV and 5% at 10 MV). The origins of these discrepancies were quantified and the shortcomings of these MU calculation methodologies are discussed in the paper.