A comparison of planned radiotherapy dose distributions calculated by Monte Carlo with variation of technique, statistical uncertainty and voxel size on resulting gamma pass rates

Background: The gamma analysis index (γ) is the most commonly used method to compare a calculated and measured dose distribution. A caveat to the clinical use of γ is that the formalism does have a sensitivity to noise present in either distribution. Monte Carlo (MC) based dose calculation methods are widely accepted as the most accurate method to calculate a resulting patient dose distribution from a radiation therapy treatment plan [6]. However, noise is inherently present as random errors or statistical uncertainty within all MC based dose calculation methods and is inversely proportional to the dose calculation time [7]. A research experiment performed by Van Delinder et al. investigated the effect of decreasing voxel size and increasing statistical uncertainty for a Monte Carlo based dose calculation method using 10 clinical head and neck (H&N) IMRT treatment plans. The experimental result was a definitive increase in γ passing rates with combined decrease in voxel size [10]. In order to further clinical information regarding this phenomenon, a large comprehensive study is required using multiple treatment techniques, a different treatment site, and using a different type of radiation measurement device. Methods/Design: A study consisting of (ntotal = 30) thirty total prostate cancer radiation therapy plans comprised of (n1 = 10) ten VMAT plans, (n2 = 10) ten IMRT plans and (n3 = 10) ten 3DCRT plans. All treatment plans consisting of three different treatment techniques are to be delivered with a total dose of 79.2 Gy prescribed to the PTV at a rate of 1.8 Gy in 44 fractions. 3DCRT will be delivered using 6 fields, IMRT delivered as 7 fields and VMAT consists of 2 arcs (CCW from 1o to 359o with a collimator angle of 170o and CW from 359o to 1o with a collimator angle of 190o). All three treatment techniques will be calculated using 6 and 10 MV energy to allow an intra-study comparison between energies. All QA plans will then be calculated with varied statistical uncertainty from 0.5%, 1%, 2%, 3%, 4% and 5%. The voxel sizes will also be varied from 3mm, 2mm and 1mm for each of the statistical uncertainty percentages. All treatment plans will be measured using ArcCHECK radiation dose measurement device and the γ will be applied within SNC Machine. Both 3%/3mm and 2%/2mm will be implemented for γ criteria for all of the treatment plans. Discussion: The necessity for a comprehensive research experiment investigating the effects of statistical uncertainty and voxel size amongst multiple treatment techniques with additional variations is required to guarantee safe quality assurance with routine use of Monte Carlo dose calculation methods. This experimental design acts as a contrasting research experiment when combined with prior literature allowing the clinician the ability to use the direct results and develop clinical recommendations for IMRT QA involving the use of Monte Carlo based dose calculations.