Absorbed fractions in the revised MIRD head phantom calculated using MCNPX

1032 Objectives MIRD Pamphlet 15 published a revised head phantom for absorbed fraction and S-value calculations. The calculations were carried out with the EGS code. To the best of our knowledge, they have not been compared with other codes. In this study, we implemented the phantom with MCNPX and compared the absorbed fractions from the two codes. Methods The phantom is a stylized model created with mathematical surfaces. Except for the back of the head and the thyroid, the MIRD surfaces are supported by MCNPX. For the unsupported surfaces, we approximated each with a series of right elliptical cylinders. A simple body was included to account for backscattering. The photon source was uniformly distributed in each source organ and it was visually cross-checked by plotting 1000 of their positions. When the target organ and the source organ were different, we used the F6 track length estimator to obtain the energy deposition; when they were the same, the *F8 pulse height tally was used. The photons and electrons were tracked until they were below their MCNPX default cut-off energies. Results The relative errors of the absorbed fractions from the simulations were generally below 1 %. For most of the values, the differences between this work and MIRD vary from less than 1 % to about 15 %. There are several target/source combinations showing significant differences, up to 60 % in the worst case. These large discrepancies were not affected by the tally choice nor by the energy indexing scheme. It could not be explained by the small difference in the organ volumes either. Conclusions MCNPX is a well-validated Monte Carlo code in medical physics applications. For most of the target/source combinations, the MCNPX calculated absorbed fractions are comparable to the MIRD ones. However, we found significant discrepancies in some cases. These large discrepancies warrant further investigation.