Monte Carlo Techniques in Diagnostic and Therapeutic Nuclear Medicine

The use of the Monte Carlo method to simulate radiation transport has become the most accurate means of predicting absorbed dose distributions and other quantities of interest in radiation treatments of cancer patients using either external or radionuclide radiotherapy. This trend has continued for the estimation of the absorbed dose in diagnostic procedures using radionuclides as well as the assessment of image quality and the quantitative accuracy of radionuclide imaging. As a consequence of this generalized use, many questions are being raised, primarily about the need and potential of Monte Carlo techniques, but also about how accurate they really are, what would it take to apply them clinically and make them available widely to the nuclear medicine community at large. Many of these questions will be answered when Monte Carlo techniques are implemented and used for more routine calculations and for in-depth investigations. The conceptual role of the Monte Carlo method is briefly introduced in this paper, which is followed by a survey of its different applications in diagnostic and therapeutic nuclear medicine. Please note that due to limited space, the references contained herein are for illustrative purposes and are not inclusive; no implication that those chosen are better than others not mentioned is intended. 1. CONCEPTUAL ROLE OF MONTE CARLO SIMULATIONS The Monte Carlo method describes a very broad area of science, in which many processes, physical systems and phenomena are simulated by statistical methods employing random numbers. The general idea of Monte Carlo analysis is to create a model, which is as similar as possible to the real physical system of interest, and to create interactions within that system based on known probabilities of occurrence, with random sampling of the probability density functions (PDFs). As the number of individual events (called histories) is increased, the quality of the reported average behaviour of the system improves, meaning that the statistical uncertainty decreases. Almost any complex system can in principle be modelled; perhaps there is a desire to model the number of cars passing a particular intersection during certain times of the day, to optimize traffic management, or to model the