RESUMO
Purpose: The aim of this study was to examine measured and calculated dose distributions in a thin-chest-wall phantom and estimate the variations in the dose-volume histogram (DVH) parameters used in plan evaluation for patient geometries with chest-wall thicknesses <15 mm with and without bolus implementation. Methods and Materials: Measurements were made using thermoluminescent dosimeters in a chest-wall phantom. The Monte Carlo method, anisotropic analytical algorithm, and Acuros XB Eclipse algorithms were used to calculate dose distributions for clinical plans. DVH parameters for clinical target volume tumor (CTVT) and planning target volume (PTV) and mean doses were evaluated for 15 patients with a chest-wall thickness of 8 to 15 mm with and without partial bolus and for 10 patients with a chest-wall thickness of 20 to 25 mm without bolus. Results: Measurements showed that the dose at a depth of 2 to 12 mm at the beam entrance and laterally was within 90% of the dose at 8 mm depth. Monte Carlo and Acuros XB calculations were well aligned with the experimental data, whereas the anisotropic analytical algorithm underestimated the beam entrance and lateral doses. The DVH parameters for the patients with a thin chest wall were sensitive to calculation algorithm, resolution, body structure definition, and patient geometry. The parameters CTVTV95%, CTVTD98%, and PTVD98% were much lower than the tolerance criteria. Partial bolus improved the values for all algorithms and decreased the variations due to patient geometry. Dose calculations for patients with a chest-wall thickness of 20 to 25 mm resulted in sufficient target coverage and low dependence on patient geometry and calculation algorithm without the use of bolus. Conclusions: Dose calculations using advanced algorithms and resolution <2 mm are recommended for patients with a thin chest wall. Specific DVH criteria or the implementation of partial bolus was needed to facilitate plan development and evaluation for this patient group.
RESUMO
PURPOSE: The aim of this study is to examine experimentally and by the Monte Carlo method the accuracy of the Eclipse Pencil Beam Convolution (PBC) and Analytical Anisotropic Algorithm (AAA) algorithms in the superficial region (0-2 cm) of the breast for tangential photon beams in a phantom case as well as in a number of patient geometries. The aim is also to identify differences in how the patient computer tomography data are handled by the treatment planning system and in the Monte Carlo simulations in order to reduce influences of these effects on the evaluation. MATERIALS AND METHODS: Measurements by thermoluminescent dosimeters and gafchromic film are performed for six MV tangential irradiation of the cylindrical solid water phantom. Tangential treatment of seven patients is investigated considering open beams. Dose distributions are obtained by the Eclipse PBC and AAA algorithms. Monte Carlo calculations are carried out by BEAMnrc/DOSXYZnrc code package. Calculations are performed with a calculation grid of 1.25×1.25×5 mm(3) for PBC and 2×2×5 mm(3) for AAA and Monte Carlo, respectively. Dose comparison is performed in both dose and spatial domains by the normalized dose difference method. RESULTS: Experimental profiles from the surface toward the geometrical center of the cylindrical phantom are obtained at the beam entrance and exit as well as laterally. Full dose is received beyond 2 mm in the lateral superficial region and beyond 7 mm at the beam entrance. Good agreement between experimental, Monte Carlo and AAA data is obtained, whereas PBC is seen to underestimate the entrance dose the first 3-4 mm and the lateral dose by more than 5% up to 8 mm depth. In the patient cases considered, AAA and Monte Carlo show agreement within 3% dose and 4 mm spatial tolerance. PBC systematically underestimates the dose at the breast apex. The dimensions of region out of tolerance vary with the local breast shape. Different interpretations of patient boundaries in Monte Carlo and the Eclipse are found to influence the evaluation. Computer tomography marker wire may introduce local disturbance effects on the comparison as well. These factors are not related to the accuracy of the calculation algorithms and their effect is taken into account in the evaluation. CONCLUSIONS: The accuracy of AAA in the case of the solid water phantom is comparable with that of the Monte Carlo method. The AAA-Monte Carlo differences in the patient cases considered are within 3%, 4 mm tolerance. The PBC algorithm does not give equivalent results. In the phantom case, PBC underestimates the lateral dose by more than 5% up to 8 mm depth. The PBC-Monte Carlo differences in the patient cases are outside the tolerance at the breast apex. The dimension of region varies with the breast shape being typically 8-10 mm long and 6-8 mm deep.
