Program Information
Electron Cutout Output Factor Prediction Model Using a Convolution Method for Highly Irregular Shaped Cutouts
J Chang*, M Lin , M Chen , S Jiang , W Lu , UT Southwestern Medical Center, Dallas, TX
Presentations
TU-RPM-GePD-T-2 (Tuesday, August 1, 2017) 3:45 PM - 4:15 PM Room: Therapy ePoster Lounge
Purpose: Electron cutout output factor, which determines the final monitor unit of electron treatment, mainly relies on measurement, especially for highly irregular cutouts. We propose a new cutout output factor and electron fluence prediction model using a convolution method.
Methods: The radiation impinging on any point on the fluence plane was modeled as an integration from all source particles in a given treatment field. Under the assumption of radially isotropic contributions, the 2D relative cutout output distribution was calculated by a convolution of a cutout shape and the convolution kernel derived from point dose measurements at the center of the fields through a series of circular fields of various sizes. This model was verified with ion chamber measurements, EDR2 film, and electron Monte Carlo (eMC) calculation from a treatment planning system at nominal and extended SSD and tested by ten clinically challenging cutout shapes.
Results: The mean absolute errors between the predicted cutout output and measurements were 0.47% (-0.66% to 1.19%), 0.61% (-1.02% to 1.08%), and 0.84 (-1.85% to 1.31%), while the errors between eMC and measurements were 0.92% (-1.39% to 1.81%), 1.17% (-1.39% to 2.3%), and 2.02% (-4.2% to 0.3%) for 6 MeV, 9 MeV, and 15 MeV, respectively. The predicted 2D cutout output distribution achieved a 97.11% mean gamma passing rate (3%/3mm) comparing to film measurements. It was also observed that the errors of the relative output distribution using the convolution method correlated with the shape complexity (perimeter²/area ratio) with an R² value of 0.872 using a linear regression.
Conclusion: We have developed an efficient and accurate electron cutout output factor and fluence map prediction model for arbitrary-shaped cutouts. This convolution-based model allows efficient calculation for the output at any given point as well as the entire 2D distributions.
Contact Email: