CCD imaging for optical tomography of gel radiation dosimeters

Several investigations have been carried out by a number of researchers over the past few years to evaluate the utility of imaginggeldosimeters for the three-dimensional measurement of radiation fields. These have been proposed to be of particular value in mapping radiationdose distributions associated with emerging and complex approaches to cancer treatment such as conformal (CRT), intensity modulated (IMRT), “gamma knife,” and pencil beam radiotherapies. Imaging of the gels has been successfully accomplished with clinical MRI units and via laser-based optical scanning. However, neither of these methods is generally accessible to all potential users, limiting the broader study and implementation of this valuable tool. We report here the design, methodology, and results of a preliminary study carried out to evaluate the utility of a new, inexpensive, and simplified approach to tomographic imaging of gelradiationdosimeters. For the purpose of this initial investigation, an array of liquid scintillation vials was prepared, containing a ferrous sulphate xylenol orange (FSX) gelatin formulation. The FSX formulation undergoes a change in optical absorption characteristics following irradiation, and the resulting color change can be observed visually. The vials were irradiated individually to different doses. Three-dimensional imaging was accomplished by tomographic reconstruction from two-dimensional optical images acquired using a diffuse, fluorescent light source, a digital charge-coupled device camera, single-photon-emission-computed tomography software, and other simple components designed by the authors. The resulting transverse images were evaluated through a region-of-interest (ROI) analysis to obtain the average change in image density in each vial as a function of radiationdose. These measured ROI values were subjected to a linear regression analysis to fit them to a straight line, and to determine the goodness of fit. Results from multiple imaging trials are compared. The correlation coefficients obtained are typically on the order of 0.98, and the p value from analysis of variance is approximately 0.05, indicating a linear and reproducible response for the dosimeter formulation and imaging system.