MABDOSE. II: Validation of a general purpose dose estimation code

The MABDOSE software represents a general tool to assess internal radiation dose. A suite of tests are described that validate the dosimetry system's implementation.

The validation suite is divided among tests that verify target digitalization, tumor digitalization and organ replacement, cumulated activity calculation, random number generation, radiation transport, and dose calculation.

A comparison between Reference Man organ volumes and MABDOSE organ volumes at (5 mm)3 resolution demonstrates volume correspondence within 10% save for ten organs having dimensions smaller than the target lattice resolution. An accounting of normal organ volume replaced by an arbitrary tumor volume indicates mass is conserved. A comparison between cumulated activities generated by MABDOSE and solutions obtained analytically demonstrates exact correspondence for curve-fitting algorithms. For mathematical modeling algorithms, cumulated activity solutions converge to their correct values provided sufficient data of high precision are input, accompanied by reasonable initial estimates of rate constants. A comparison of MABDOSE results with the MIRD 3 report demonstrates good agreement (<8% difference) in absorbed fractions for spheres at energies from 20 keV to 2.75 MeV. A comparison of MABDOSE results with the Cristy-Eckerman report demonstrates marginal agreement (specific absorbed fractions within a factor of 2 for all Reference Man organs) at simulation energies of 20, 50, and 100 keV. Lack of exact correspondence is attributed to volume digitalization errors, and to differences in cross-section libraries, interpolation schemes between cross-section data points, and random number generators. Finally, the doses reported by MABDOSE correspond to the correct algebraic combination of paired cumulated activities and "S" values.

The MABDOSE program has been validated as a general purpose computation tool for use in internal radionuclide dosimetry.