A new 3-dimensional (3-D) computation algorithm, "Modified equivalent TMR method", for photon beam dose calculation, considering heterogeneous corrections, has been developed. This algorithm is base on the modified equivalent TAR method that was proposed by Inamura et al. The modified equivalent TAR method was deduced from the equivalent TAR method proposed originally by Sontag and Cunning ham. In addition, reducing the computational time during treatment planning, the concept of quasirandom number suggested by Inamura et al, is adopted for this algorithm. The purpose of this study is to evaluate the accuracy of the dose computations of this new algorithm. The accuracy is compared with both measure and calculated data for Cobalt-60 gamma rays and 10 MV photon beams for various geometries and field sizes on the axis. Weighting factor expressing the relative importance of each volume element in contributing to the scatter dose at the point of calculation, is computed using the three following models that are exponentially approximated as the distance function from the scattering point to the calculation one. 1) "All Scatter model" accounts for the all scattering points. 2) "No Back Scatter model" only takes account of the forward scattered radiation at the calculation point. 3) "Compton Scatter model" takes account for the contributing rate of the scatter dose at the calculating point using the approximation of Klein-Nishina's angular distribution. Using the modified equivalent TMR method, we show that the accuracy of a 3-D dose computation is less than 2% for a homogeneous medium and about 6% for a heterogeneous one. This algorithm is possible for calculating a 3-D dose distribution for short periods of time in actual treatment planning and also for to photon beam applications with more than 6 MV.
