LAUSR

Abstract Purpose The Leksell Gamma Plan Convolution algorithm (LGP‐Convolution) has not been widely adopted. This mainly stems from the higher calculated beam‐on times relative to the standard ray tracing‐based LGP‐TMR10 dose calculation algorithm. This study aims to evaluate the accuracy of the LGP‐Convolution in scenarios where the treated lesions are in the vicinity of or encompassed by bone and/or air inhomogeneities. Methods The solid water dosimetry phantom provided by the vendor was modified to include bone and air inhomogeneities. Two treatment planning scenarios were investigated involving a single shot and multiple shots, respectively. Treatment planning and dose prescription were performed using the LGP‐Convolution algorithm. Triple channel film dosimetry was performed using GafChromic EBT3 films calibrated in terms of absorbed dose to water in a 60Co beam. Monte Carlo (MC) simulation dosimetry was also performed in the inhomogeneous experimental geometry using the EGSnrc MC platform and a previously validated sector‐based phase‐space source model. MC simulations were also employed to determine correction factors required for converting EBT3 measurements at points within the bone and air inhomogeneities from dose‐to‐water values to the corresponding dose to medium values. Results and Conclusions EBT3 dose to medium correction factors ranged with field size (4, 8, or 16 mm) within 0.941–0.946 for bone and 0.745–0.749 for air inhomogeneities. An excellent agreement was found between the LGP‐Convolution calculations with corresponding EBT3 and MC dose to medium results at all measurement points, except those located inside the air inhomogeneity. The latter is of no clinical importance and excluding them yielded gamma index passing rates of nearly 100% for 3% local dose difference and 1 mm distance‐to‐agreement criteria. The excellent agreement observed between LGP‐Convolution calculations and film as well as MC results of dose to medium indicates that the latter is the quantity reported by the LGP‐Convolution.