Reconstruction of high-energy bremsstrahlung spectra by numerical analysis of depth-dose data

The determination of the spectral distribution of a photon beam is important for exact dosimetry and for the calculation of dose distributions. Direct measurements are not possible because of physical reasons. The approximation described uses a numerical reconstruction method based upon depth-dose data measured in water. The basic idea of the method is that a measured depth-dose curve can be thought of as a weighted superposition of monoenergetic depth-dose curves. The monoenergetic data are obtained by convolution of Monte Carlo generated energy deposition kernels. They were fitted to the measured data by means of a least squares method. Several constraints were applied to the superposition coefficients. The optimization algorithm uses the Wolfe reduced gradient method. The results were compared to Monte Carlo generated thick target spectra.