Derivation of photon energy spectra from transmission measurements using large fields

Modern treatment planning systems based on Monte Carlo technique require, in order to calculate the dose, knowledge of the photon spectra produced by medical linear accelerators. The accuracy of the dose determination will increase when the spectra are better known.; In the present work the 6 MV photon spectrum of a Varian 2100C linear accelerator was determined from attenuation measurements performed in large fields. The iterative algorithm written in MathematicaRTM used as input data Monte Carlo-predetermined pencil beam monoenergetic scatter kernels for various water phantom thicknesses, open beam fluences and beam fluences measured in air with phantoms of different thicknesses placed in the beam. The experimental data was measured using an ionization chamber and two types of film, GAFCHROMICRTMEBT film and KODAK EDR2 film. The iteration started with a flat spectrum used to calculate the polyenergetic kernels for each water thickness. The spectrum-dependent scatter for different thicknesses of water was calculated convolving the corresponding polyenergetic kernel with the signal obtained with the water phantom removed from the beam. For each thickness of water, transmissions on the central axis were given by the ratios of central axis primary fluences to the open beam fluence. The reconstructed energy spectrum was determined from the transmission values using the simulated annealing technique. Simulated annealing was preferred because it reaches the true global minimum better than other optimization techniques. The spectrum determined at the end of the simulated annealing loop was compared to the input spectrum of the general algorithm. If they matched within acceptable errors this was the final primary spectrum. If not, the spectrum was fed as input for a new iteration.; Monte Carlo monoenergetic scatter kernels were derived for six water thicknesses. The amplitude of the monoenergetic scatter kernels increases with energy and water phantom thickness. For thin phantoms there is a strong dependence of scatter with thickness. For large phantoms the increase is negligible after a certain phantom thickness which depends on beam energy. The average energy of the reconstructed spectrum was 2 MeV. The accuracy of the spectrum was checked by comparing MCNP-computed percent depth dose with the direct measurement of this quantity. There was a good agreement (2%) between measured and calculated PDD except at the first portion of the graph. The discrepancy could be due to the electron contamination from the head of the linear accelerator, which was not taken into account in this algorithm.; Modeling of the head of linear accelerator is a time consuming task. The presented algorithm has the advantages that by-pass the tedious step of modeling the head of linear accelerator these effects being already included in the measured fluences. The method is robust, good for portal dosimetry. It can be used to evaluate accurately the photon scatter in portal imaging since is taking into account the energy spectrum dependence of the scatter, while other papers assume known spectrum.