Abstract
49
Objectives The development of voxel based Monte Carlo methods has enabled patient-specific dose calculations for targeted radionuclide therapy. A rate-limiting step in the implementation of this technique is the need to define the anatomical data of the patient; Monte Carlo transport calculations are now possible in reasonable times through the use of computing clusters. We propose a novel, quick method for creating patient-specific voxelized phantoms using deformable anthropomorphic NURBS models as input for Monte Carlo dose calculations using the Geant4 simulation toolkit.
Methods We have developed a toolkit to deform standard phantoms to reflect patient anatomy. The phantom anatomy is matched to CT data using a variety of methods that manipulate deformable (NURBS-based) surfaces defining the phantom body and organs. The voxelized patient-specific phantom is then paired with activity distribution data from quantitative SPECT or PET as input into our Geant4 simulation code that models radionuclide decay, transporting both photons and electrons, to obtain 3D absorbed dose maps.
Results We evaluated organ and tumor doses for several I-131 radioimmunotherapy patients. The patient-specific NURBS models were created in 1-2 hours and had organ masses within 10% of those obtained by traditional segmentation methods. Voxelized dose calculations using Monte Carlo simulation are completed in several hours, with relative errors in each organ less than 5%. Organ doses were compared to results from established dosimetry codes (e.g. OLINDA/EXM) and dose maps and DVHs were compared to results reported by others (e.g. adapted Dose Planning Method).
Conclusions The application of this technique to several patient studies demonstrates the use of deformable phantoms as an accurate and efficient method for creating patient-specific anatomical models. These models, used as input to Monte Carlo simulations for dose calculations, allow 3D dosimetry for targeted radionuclide therapy to be performed in times that are reasonable for clinical use.