Abstract

The use of radiolabeled molecules for tumor targeting constitutes a remarkable technique for the treatment of systemic malignancies. An accurate patient-specific dosimetry in nuclear medicine procedures should be a relevant pre-requisite in order to achieve the required lethal damage to tumor cells while maintaining possible side-effects to normal tissues at tolerable levels. It is desired to assess in vivo the radiopharmaceutical distribution for further estimation of absorbed dose released to target and involved organs. In this context, it was developed a computational toolkit, called DOSIS, in order to perform patient-specific dosimetry based on personalized patient anatomy and biodistribution of radionuclides both obtained by currently available dual PET/CT or SPECT/CT facilities. This work is focused on comparing 3D dose distributions obtained by DOSIS performing full stochastic Monte Carlo simulations versus analogue distributions obtained with analytical approaches like dose point kernel convolution and local energy deposition, when considering non-homogeneous activity or density distributions at different scales. Mathematical virtual phantoms were created for this study in order to compare results with other calculation methods. Some of the beta-emitters radionuclides commonly used for therapy (Y-90, I-131, Lu-177) were investigated, and emissions of beta-particles, conversion electrons, gamma radiation, and characteristic X-rays were considered. DOSIS implements a novel code devoted to managing radiation transport simulation by means of PENELOPE Monte Carlo general-purpose routines on voxelized geometries defined by 3D mass and activity distributions. Both distributions can be defined through patients-specific images, or pre-defined virtual phantoms. Results preliminary confirmed DOSIS as a reliable and accurate toolkit for personalized internal dosimetry along with highlighting advantages/drawbacks of the different calculation schemes proposed.