Influence of CyberKnife Prescription Isodose Line on the Discrepancy of Dose Results Calculated by the Ray Tracing and Monte Carlo Algorithms for Head and Lung Plans: A Phantom Study.

Abstract

Incorporation of the Monte Carlo (MC) algorithm in optimizing CyberKnife (CK) plans is cumbersome, and early models unconfgured MC calculations, therefore, this study investigated algorithm-based dose calculation discrepancies by selecting different prescription isodose lines (PIDLs) in head and lung CK plans. CK plans were based on anthropomorphic phantoms. Four shells were set at 2-60 mm from the target, and the constraint doses were adjusted according to the design strategy. After optimization, 30%-90% PIDL plans were generated by ray tracing (RT). In the evaluation module, CK plans were recalculated using the MC algorithm. Therefore, the dosimetric parameters of different PIDL plans based on the RT and MC algorithms were obtained and analyzed. The discrepancies (mean±SD) were 3.72%±0.31%, 3.40%±0.11%, 3.47%±0.32%, 0.17%±0.11%, 0.64%±3.60%, 7.73%±1.60%, 14.62%±3.21% and 10.10%±1.57% for D1%, D(mean), D98% and coverage of the PTV, DGI, V5, V3 and V1 in the head plans and -6.32%±1.15%, -13.46%±0.98%, -20.63%±2.25%, -34.78%±25.03%, 122.48%±175.60%, -12.92%±5.41%, 3.19%±4.67% and 7.13%±1.56% in the lung plans, respectively. The following parameters were signifcantly correlated with PIDL: dD98% at the 0.05 level and dDGI, dV5 and dV3 at the 0.01 level for the head plans; dD98% at the 0.05 level and dD1%, dD(mean), dCoverage, dDGI, dV5 and dV3 at the 0.01 level for the lung plans. RT may be used to calculate the dose in CK head plans, but when the dose of organs at risk is close to the limit, it is necessary to refer to the MC results or to further optimize the CK plan to reduce the dose. For lung plans, the MC algorithm is recommended. For early models without the MC algorithm, a lower PIDL plan is recommended; otherwise, a large PIDL plan risks serious underdosage in the target area.