A fast voxel-based Monte Carlo method for scanner- and patient-specific dose calculations in computed tomography

It is difficult to determine the patient dose caused by CT examinations from the manufacturer's technical specifications of the scanner, because with these the scanner is only characterized roughly and the geometry of the patient is not taken into account at all. To remedy this lack of information concerning the patient dose we have developed a fast voxel-based Monte Carlo program (ImpactMC) to calculate dose distributions specifically for both the respective CT scanner and the individual patient anatomy. The calculations are performed on a 3-D voxel volume that can be filled with either the CT data of patients or mathematical phantoms converted to voxel volume data. In addition information about the scanner, e.g., scanner geometry, pre-filtration and shaped filters, and about the scanning parameters, e.g., tube current, slice thickness and pitch, has to be provided. According to these parameters the dose distribution in the 3-D voxel volume is then calculated by the Monte Carlo method. Several variance reduction techniques were implemented to minimize the time for calculation. To validate the program, the calculated results were compared with the results obtained with Monte Carlo-based programs (EGS4), measured CTDI values and published values of organ dose calculations. Typically, the differences were less than 5%. The time for calculation depends on the capability of the PC system, the 3-D volume and the simulated problem. E.g. the effective dose for a CT examination of the liver is calculated in less than three minutes with an uncertainty of less than 1% standard deviation. Moreover, the program allows the quantification of the potential of dose reduction techniques and the optimization of the respective method - specific to the scanner and the patient.