Abstract
X-ray film emulsion contains high-Z elements that preferentially respond to photons with energies below 400 keV. This over-response significantly overestimates the dose at deep parts of a phantom where the low-energy photons are abundant due to scattering. Previous method used a high Z foil placed next to film to improve the dose response. However, this method cannot be used for off-axis dose measurement. It’s inappropriate for multiple fields dose verification.We propose an improved film system by using the combination of film with various filters (None, Pb, Cu, and Al) so that each film is irradiated to different energy spectrum and the readings of these four films are input to a neural net that relates the actual dose with energy spectra.Percentage depth dose (PDD) measured by the film system in the polystyrene phantom for a 10cm×10cm field size of a 6MV linear accelerator together with measurement in a water phantom using an ionization chamber detector were served as a training data set to the neural net. To validate this method, the PDD of a 20×20 field size and the dose distribution of an off-axis location at 5cm are estimated using this trained neural network. The maximum differences between the measurements of film system and ionization chamber were less than ±4%, and less than 2% for the mean error in our investigation.Further study of this film system will focus on the performance of multi-fields and its clinical application.