Evaluation of an electron Monte Carlo dose calculation algorithm for treatment planning

Abstract

The purpose of this study is to evaluate the accuracy of the electron Monte Carlo (eMC) dose calculation algorithm included in a commercial treatment planning system and compare its performance against an electron pencil beam algorithm. Several tests were performed to explore the system's behavior in simple geometries and in configurations encountered in clinical practice. The first series of tests were executed in a homogeneous water phantom, where experimental measurements and eMC-calculated dose distributions were compared for various combinations of energy and applicator. More specifically, we compared beam profiles and depth-dose curves at different source-to-surface distances (SSDs) and gantry angles, by using dose difference and distance to agreement. Also, we compared output factors, we studied the effects of algorithm input parameters, which are the random number generator seed, as well as the calculation grid size, and we performed a calculation time evaluation. Three different inhomogeneous solid phantoms were built, using high- and low-density materials inserts, to clinically simulate relevant heterogeneity conditions: a small air cylinder within a homogeneous phantom, a lung phantom, and a chest wall phantom. We also used an anthropomorphic phantom to perform comparison of eMC calculations to measurements. Finally, we proceeded with an evaluation of the eMC algorithm on a clinical case of nose cancer. In all mentioned cases, measurements, carried out by means of XV-2 films, radiographic films or EBT2 Gafchromic films. were used to compare eMC calculations with dose distributions obtained from an electron pencil beam algorithm. eMC calculations in the water phantom were accurate. Discrepancies for depth-dose curves and beam profiles were under 2.5% and 2 mm. Dose calculations with eMC for the small air cylinder and the lung phantom agreed within 2% and 4%, respectively. eMC calculations for the chest wall phantom and the anthropomorphic phantom also showed a positive agreement with the measurements. The retrospective dosimetric comparison of a clinical case, which presented scatter perturbations by air cavities, showed a difference in dose of up to 20% between pencil beam and eMC algorithms. When comparing to the pencil beam algorithm, eMC calculations are definitely more accurate at predicting large dose perturbations due to inhomogeneities.

Figure 1

Phantom simulating a chest wall.

Figure 2

CT slice of the thorax part of RANDO.

Figure 3

Seed number effect in water for a $6\times 6\hspace{0.17em}{\text{cm}}^2$ field size at 18 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$. Ten calculations with different seed numbers are averaged and a dispersion of $\pm 2\sigma$ is also calculated. The measurement was done with an ionization chamber.

Figure 4

Depth‐dose curves in water for different resolution parameters for a $6\times 6\hspace{0.17em}{\text{cm}}^2$ field size at 6 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$. The measurement was done with an ionization chamber.

Figure 5

Measured and calculated crossbeam profiles in water at different depths (1.9 cm, 3.7 cm, and 4.9 cm) for a $10\times 10\hspace{0.17em}{\text{cm}}^2$ field size at 12 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$. Measurements were performed with an IC10 ionization chamber.

Figure 6

Measured and calculated crossbeam profiles in water at different depths (1.9 cm, 3.7 cm, and 4.9 cm) for a $10\times 10\hspace{0.17em}{\text{cm}}^2$ field size at 12 MeV and $\text{SSD}=110\hspace{0.17em}\text{cm}$. Measurements were performed with an ionization chamber. All beam profiles were normalized according to the measured and calculated depth‐dose curve.

Figure 7

Measured and calculated crossbeam profiles in water at different depths ($1.9\hspace{0.17em}\text{cm},3.7\hspace{0.17em}\text{cm}$, and $4.9\hspace{0.17em}\text{cm}$) for a $10\times 10\hspace{0.17em}{\text{cm}}^2$ field size, at 12 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$. Gantry angles $\text{GA}={20}^{\circ }$. Measurements were performed with an ionization chamber. All beam profiles were normalized according to the measured and calculated depth‐dose curve.

Figure 8

Crossbeam profiles in the air cylinder phantom for a $10\times 10\hspace{0.17em}{\text{cm}}^2$ applicator at 12 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$: (a) profile at the air cylinder–plastic water interface (2.7 cm of depth); (b) profiles in plastic water at two different depths (4.2 cm and 6.3 cm); (c) PWDT‐corrected profiles in plastic water (4.2 cm and 6.3 cm).

Figure 9

Crossbeam profiles in the first version of the lung phantom and a depth‐dose curve in the second version of the lung phantom (12 MeV, $10\times 10\hspace{0.17em}{\text{cm}}^2$, $\text{SSD}=100\hspace{0.17em}\text{cm}$): (a) profile at the plastic water–cork interface (2 cm of depth); (b) profiles in the cork at different depths (3.6 cm and 8.4 cm); (c) depth‐dose curve (0–2 cm water, 2–11 cm cork).

Figure 10

Crossbeam profiles in the chest wall phantom for a $20\times 20\hspace{0.17em}{\text{cm}}^2$ applicator at 12 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$: (a) profile at 2.7 cm of depth; (b) profile at 4.3 cm of depth; (c) profile at 2.7 cm of depth, with the stopping power correction for the Teflon part; (d) profile at 4.3 cm of depth, with the stopping power correction for the Teflon part.

Figure 11

Illustration of superposed measured and calculated absolute isodoses in RANDO for a $15\times 15\hspace{0.17em}{\text{cm}}^2$ applicator at 18 MeV and $\text{SSD}=100\hspace{0.17em}\text{cm}$: (a) pencil beam algorithm vs. EBT2 film; (b) eMC algorithm vs. EBT2 film.

Figure 12

Illustration (a) of the absolute isodoses from the pencil beam algorithm on a CT slice of the clinical case (doses in Gy: 1.5, 3, 10, 18, 30, 40, 51, 57, 60, 63). Illustration of the absolute isodoses from the eMC algorithm (b) on the same CT slice of (a). The planned target (PTV) volume is defined by the green shadowed region and the clinical target volume (CTV) is defined by the orange shadowed region. In (a) and (b), the prescription dose is 60 Gy and 210 MU were delivered for both algorithms. Map (c) of dose differences as calculated by the two planning systems (pencil beam minus eMC). DVH graph (d) for the PTV and the CTV for both algorithms.