A method for deconvolution of integrated electronic portal images to obtain incident fluence for dose reconstruction

A method to convert integrated electronic portal imaging device (EPID) images to fluence for the purpose of reconstructing the dose to a phantom is investigated here for simple open fields. Ultimately, the goal is to develop a method to reconstruct the dose to patients. The EPID images are transformed into incident intensity fluence by spatial filtering with a deconvolution kernel. The kernel uses a general mathematical form derived from a Monte Carlo calculation of the point spread function of an EPID. The deconvolution kernel is fitted using a downhill search algorithm that minimizes the difference between the reconstructed dose and the dose measured in water. The beam profile "horns" that are removed by the EPID calibration procedure are restored to the resulting images by direct multiplication using the measured in-air off-axis ratio. Applying the fitted kernel to an EPID image provides the incident fluence for that beam. This beam fluence is then entered into an independent dose calculation algorithm for phantom or patient dose reconstruction. The phantom dose was computed to an accuracy of 2.0% of the dmax dose at one standard deviation. The method is general and can possibly be applied to any EPID equipped with an integration mode. We demonstrate the application of the fitted kernel in two clinical IMRT cases.