Aim: Targeted radionuclide therapy applications require the use of small animals for preclinical experiments. Accurate dose estimation is needed in such animals to explore and analyze the toxicity of injected radiopharmaceuticals. We developed two numerical models to allow for a more accurate mouse dosimetry.
Methods: A frozen nude mouse (30 g) was sliced and digital photographs were taken during the operation. More than 30 organs and tissues were identified and manually segmented. A digital (voxel-based) and a mathematical model were constructed from the segmented images. Important organs were simulated as radiation sources using the Monte-Carlo code MCNP4C. Mono-energetic photons from 0.005 to 2 MeV, and monoenergetic electrons from 0.1 to 2.5 MeV were simulated. Activity was supposed to be uniform in all source organs.
Results: Results from monoenergetic emissions were integrated over emission spectra. Radionuclide S-factors (Gy/Bq.s) were calculated by taking into account both electron and photon contributions. A comparison of the results obtained with either a voxel-based or mathematical model was carried out. The voxel-based model was then used to revise dosimetric results, obtained previously under the assumption that all emitted energy was absorbed locally. For (188)Re, the self-absorbed doses in xenografted tumors were 39-69% lower than that obtained by assuming local energy deposition.
Conclusion: The voxel-based models represent more realistic anatomic approach. The rapid advancement of computer science and new features added to Monte-Carlo codes permit considerable reduction of computational run time. Cross-doses should not be neglected when medium to high energy beta emitters are being used for preclinical experiments on mice.