Comparison of Treatment Planning Systems with Monte Carlo Simulation Under Conditions of Tissue Inhomogeneities

This study aimed to validate a Monte Carlo (MC) beam model of a 6-MV Elekta Synergy MLCi linac and benchmark two different treatment planning systems (TPSs) employing pencil-beam (PB) and collapsed-cone convolution/superposition (CCC) algorithms against MC in heterogeneous slab phantoms. The initial electron energy was optimized by χ² analysis of water-phantom PDDs and lateral profiles for a 10×10 cm2 field (5.4–6.6 MeV); 5.8 MeV minimized χ² and was adopted. Agreement with measurement was verified using percent depth dose curves, lateral dose profiles, TPR20,10 and output factors. Dose distributions of 3×3, 10×10 and 20×20 cm2 open fields in lung-, bone- and water-equivalent virtual slab phantoms were then calculated using both TPSs and MC. MC calculations were performed in two modes: one yielding dose-to-medium (MCDm) and the other yielding dose-to-water (MCDw). At 5.8 MeV, measurements versus MC comparisons showed <0.5% differences for PDDs (≤20×20 cm²), ~1% for lateral profiles (outside high-gradient regions), TPR20,10​ <0.2% and ~1% for output factors. In the lung phantom and 3×3 cm2 field, PDD differences of up to 10% were observed between TPSPB and MCDw, whereas the corresponding difference for TPSCCC was 4.5%. The differences were smaller in the other phantoms. In the dose profile comparisons, particularly at the field edges, discrepancies of up to 14% were observed. In conclusion, the MC model demonstrated a high level of agreement with measurements. TPSCCC calculations were closer to those by MCDw than TPSPB.