Validation of a Monte Carlo prediction model for portal images using PENELOPE

Abstract : In external beam radiotherapy, dosimetric verification is usually performed by comparing dose images provided by EPIDs with predicted images computed using Monte Carlo (MC) simulations. The goal of the present study is to develop a portal prediction model with the MC code PENELOPE and validate it against experiment. Experimental images were acquired at 12 MV using a Saturne 43 accelerator and a fluoroscopic Lynx2D EPID (SDD = 150 cm), for 4´4, 10´10 and 15´15 cm² open fields, and both without and with a 30´30´30 cm3 water phantom in the beam. The accelerator and the EPID were simulated with PENELOPE. Two EPID models were implemented, one describing in details the geometry and the other being a simplified model, composed of three layers (copper, GOS and water). The water layer simulated the backscatter within the EPID and its thickness had to be adjusted. The best EPID model was found to be the simplified model including an 8 cm thick backscatter compartment. This model was then used to compute portal images with and without the water phantom in the beam and the simulated and acquired images were compared using the g-analysis. g-index values less than 1 were obtained for 96% of the pixels for fields of 10x10 cm² or less and for 90% of the pixels for larger field sizes. These results exhibit that a simple MC EPID model allows predicting accurately portal images for open fields, with and without object in the beam, and that PENELOPE is a reliable tool for this purpose.