Monte Carlo validation in diagnostic radiological imaging

Monte Carlo analysis in the radiological sciences has been used for several decades, however with the ever-increasing power of desktop computers, the utility of Monte Carlo simulation is increasing. A Monte Carlo code called the Simple Investigative Environment for Radiological Research Applications (SIERRA) is described mathematically, and is then compared against an arra y of published and unpublished results determined by other means. A series of 32 comparisons between data sets, 22 from independent Monte Carlo simula tions and 10 from physically measured data, were assessed. The compared par ameters included depth dose curves, lateral energy scattering profiles, sca tter to primary ratios, normalized glandular doses, angular scattering dist ributions, and computed tomography dose index (CTDI) values. Three of the 3 2 comparison data sets were excluded as they were identified as outliers. O f the remaining 29 data sets compared, the mean differences ranged from - 1 4.8% to + 17.2%, and the average of the mean differences was 0.12% (sigma=1 .64%), and the median difference was 1.57%. Fifty percent of the comparison s showed mean differences of similar to 5% or less, and 93% of the comparis ons showed mean differences of 12% or less. We conclude that for research a pplications in diagnostic radiology, the SIERRA Monte Carlo code demonstrat es accuracy and precision to well within acceptable levels. (C) 2000 Americ an Association of Physicists in Medicine. [S0094-2405(00)01506-6].