Comparison of video and CCD cameras in online portal imagers calibrated for dosimetry

Abstract

Video-based electronic portal imaging devices (EPIDs) are the most common type of portal imager found in clinical practice. In addition to providing a method to detect and correct patient set-up errors and record treatments, possibilities exist for the dosimetric use of video-based EPIDs. To determine if a particular system is suitable for dosimetry, the entire system and especially the camera should be thoroughly tested and calibrated. Comparisons were made of the performance of a Newvicon camera and a CCD camera in a video-based EPID. Tests were performed on the cameras to investigate linearity, noise, frame grabber effects, clamping error, and lens vignetting. Tests of the entire imaging system were performed to determine the behaviour of the veiling glare, the uniformity of the spatial sensitivity, and the EPID response with changes in field size and phantom thickness. The veiling glare was caused by several factors: the mirror, the cameras themselves, and the phantom scatter and beam hardening. The magnitude of the glare was spatially dependent. The EPID response was not spatially uniform, but increased by approximately 25% i the region farthest from the screen-mirror junction. The cause of this increase was determined to be the mirror. The dose-corrected EPID response with changes in phantom thickness was nearly flat for medium-sized fields, although it did increase slightly for large fields and decrease slightly for small fields. The overall performance of the CCD camera was found to be significantly better than that of the Newvicon camera. Central axis dosimetry would be straightforward to calibrate; however, the spatially dependent glare observed in the system would make it extremely difficult to deconvolve an EPID image to provide a two-dimensional dose map.