Modeling of human body tissue compositions for Monte Carlo algorithm of Proton therapy dose computation with the Single Energy Computed Tomography Calibration Curve

Proton dose computation with most planning systems rely on single energy computed tomography (CT) images in which the relative proton stopping-power ratio (SPR), mass density and the relative electron density are derived from CT Hounsfield units (HU). Using a proton Monte Carlo dose calculation algorithm in treatment planning system (TPS), a CT number of each pixel in patient CT image is converted to mass density; then the proton stopping power ratio is computed. The accuracy of proton dose computation in Monte Carlo algorithm relies on conversion from HU to mass density. We explore the potential improvement in determining mass density to reduce the uncertainty in predicting the proton range in patients. The Stoichiometric method is used to calculate the CT scanner specific parameters related to the photoelectric effect, coherent scattering, and Compton interactions, in order to model the CT number of human body tissue compositions. In the stoichiometric calibration curve, elemental compositions and densities of 34 “standard” human biological tissues have been taken from the ICRU report 44. Our results demonstrate that a more accurate prediction of HU to mass density can be achieved by the stoichiometric calibration curve for proton therapy dose computation (rms error 0.81%).