Room: AAPM ePoster Library
Purpose: propose a continuous dose calculation method for volumetric modulated arc therapy (VMAT) using pencil beam (PB) convolution in a homogeneous phantom, reducing dose discrepancies due to the under-sampling issue.
Methods: BEAMnrc/DOSXYZnrc Monte Carlo (MC) simulations were used to model 6 MV pencil beams (1 x 5 mm²) on a homogeneous (water) cylindrical phantom with 50-cm diameter placed at 100 cm source axis distance. MC generated PB kernels at varied depths and slices of the phantom in polar coordinate system were collected for each pixel of a 40 x 40 cm² grid at isocenter. The intensity for each pixel was determined by multileaf collimators and jaws positions from patient RTP files. Dose distributions at a desired depth and slice of the phantom can be calculated by a summation of convolutions of the corresponding PB kernels and intensity for each pixel using fast Fourier transforms. Calculated and MC simulated dose distributions in polar coordinate system were compared using gamma analysis. For dose comparison in Cartesian coordinate system, calculated dose distributions in polar coordinate system were first obtained by PB convolution, then converted to Cartesian coordinate system using interpolation. Theoretically, the computational efficiency of the proposed continuous method should be better than the discretized method since convolution is applied to cylindrical system in our method whereas convolution for discretized method needs to be done at each discretized angle.
Results: polar coordinate system, the gamma passing rate between calculated and simulated dose distributions was 100% (1%/1mm). In Cartesian coordinate system, the gamma passing rate between calculated and simulated dose distributions was 96% (1%/1mm) and 98% (2%/2mm).
Conclusion: results suggest that the proposed method can be implemented for VMAT dose calculation in a homogeneous phantom without the under-sampling issue, which can potentially increase the efficiency and improve the accuracy of VMAT dose calculation.