Monte carlo based calculation of patient exposure in x-ray ct-examinations

Abstract

The Monte Carlo (MC) method is a powerful tool to examine the impact of different CT scan-parameters, protocols and individual anatomy on patient dose. General information about dose exposition required by radiation protection legislation for CT examination is mostly based on measurements or generalized MC simulations with humanoid phantoms. This is an approximation to the actual patient anatomy and must be considered as an estimation of the actually applied dose. In special cases, such as the abdominal CT of a pregnant woman, an exact and individual dose calculation could improve the information required for decisions regarding the need for subsequent abortion. MC simulations are time consuming and the handling of available MC packages requires a skilled user. We introduce a free software tool GMctdospp which serves as a front-end and user-code for the general purpose MC-package EGSnrc. All scan parameters needed for a simulation of the CT-examination can be intuitively defined in the graphical user interface. The graphical representation and analysis of resulting dose distributions by means of profiles and dose volume histograms is included in the front-end. Further, the complete controlling of a distributed computing environment is performed by GMctdospp, allowing an arbitrary number of conventional desktop PCs to be used for calculation. CT data sets can be loaded in the medical imaging standard DICOM-format. Regions of interest (ROIs) can be defined as DICOMRT struct and evaluated statistically as mean dose or dose volume histogram. The required conversion of CT units (Hounsfield) to mass densities and materials needed for the MC simulation can be defined (defaults are set). Calculating absolute dose distributions is possible with the use of a calibration run, where dose in an ionisation chamber is once measured and then calculated for identical beam source parameters (energy spectrum, filtration, slice thickness etc.). © 2009 Springer Berlin Heidelberg.