Abstract
PURPOSE
Rapid adoption of targeted radionuclide therapy as an oncologic intervention has motivated the development of patient-specific voxel-wise approaches to radiation dosimetry. These approaches often rely on pretabulated dose point kernels for convolution-based calculations; however, these dose kernels are sparse in literature and often have suboptimal characteristics. The purpose of this work was to generate an extensive library of dose point kernels with sufficient size and resolution for general clinical application of voxel-wise dosimetry.
METHODS
Nuclear data were acquired for 2174 radionuclides from the National Nuclear Data Center (Brookhaven National Laboratory, accessed March 2018). Based on these data, isotropic point sources of radioactivity in water were simulated using Monte Carlo N-Particle transport v6.2 (MCNP6.2, Los Alamos National Laboratory). Simulations were separated by emission type for each radionuclide - photons (γ-rays, x rays), beta particles (positrons, electrons); and discrete electrons (conversion electrons, Auger electrons, Coster-Kronig electrons). Dose was tallied in concentric spherical shells about the point source using an energy deposition pulse-height tally (MCNP *F8 tally). Bins were spaced every 0.1 mm until a radius of 10 cm, and every 1 mm until a radius of 2 m. Positron emissions where treated as electrons for transport, with annihilation photons generated at the origin within the photon simulation. Alpha particle emissions were not simulated since their energy is deposited within ~0.2 mm of the source. Neutron and spallation effects were not considered. A subset of the resultant dose point kernels (11 C, 18 F, 32 P, 52g Mn, 64 Cu, 67 Ga, 89 Sr, 89 Zr, 90 Y, 99m Tc, 111 In, 117m Sn, 123 I, 124 I, 125 I, 131 I, 153 Sm, 177 Lu, 186 Re, 188 Re, 211 As, 212 Pb, 213 Bi, 223 Ra, and 225 Ac) were evaluated for accuracy based on conservation of energy, comparison to kernels in the literature, and statistical precision.
RESULTS
Among dose point kernels that were manually reviewed, good agreement with previously published dose point kernels was observed. Energy within the kernels was found to be conserved to within 1% of the value expected from nuclear data, suggesting that a radius of 2 m was sufficient to capture the almost all of the energy released during decay for all isotopes considered. Local dosimetric uncertainty, evaluated at the radius of 99% energy deposition, was found to be less than 9% for all radioisotopes evaluated. Rebinning data more coarsely by a factor of 10, similar to what would be done for a clinical dose calculation, results in all evaluated kernels having a relative error of less than 1.1% at R50% , 1.5% at R90% , and 2.7% at R99% (the radius corresponding to 50%, 90%, and 99% of total energy deposition, respectively). The kernels produced in this work have been made freely available (https://zenodo.org/record/2564036).
CONCLUSIONS
An extensive library of high-resolution radial dose kernels was generated and validated against published data. In addition to enabling patient-specific voxel-wise internal dosimetry by convolution superposition, the generated dose point kernels data may prove useful to the wider health physics community.
Collapse