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

Abstract

The use of an electronic portal imaging device (EPID) to measure beam fluence intensities to be used to reconstruct the dose to patients is investigated. The EPID is used to acquire images of the treatment fields that are to be used to treat a patient with megavoltage radiation therapy. In this study, images are obtained without the patient or couch in the beam and are integrated for the full monitor unit prescription for each treatment portal. The EPID images are transformed into intensity fluence by spatial filtering with a deconvolution kernel. The in-air off-axis ratio that was removed by the EPID calibration procedure is restored to the resulting images by direct multiplication. The deconvolution kernel is fitted using a general mathematical form for the point spread function of an EPID using a downhill search algorithm to minimize the difference between the reconstructed dose and the dose measured in water. The dose to the patient can be reconstructed given the patient?s CT data set, stated beam positions, and energy. This derived intensity fluence distribution defines the x-ray intensity over the area of the beam and includes all effects upon the intensity due to field modulation, geometric penumbra effects, attenuators placed in the beam, and beam limiting devices. The resulting intensity fluence distribution is used as the definition of the beam for input to the dose calculation algorithm. The EPID is thus used to integrate each treatment portal for the beam-on time to measure the two-dimensional beam profile for that portal. The dose that would result in a phantom or patient from the application of that treatment beam is computed. The dose at depth for the field sizes used for the kernel fit was computed to an accuracy of 2.0% of the dmax dose at one standard deviation. The method is general and can be applied to any EPID equipped with an integration mode. The application of this method will allow the dose distribution to a patient to be determined by measuring patient-specific treatment beams in the absence of the patient, providing a convenient method for treatment verification. The dose found here may then be directly compared to the planned dose.