Development and evaluation of a quantitative reconstruction method for Th-227 SPECT

Objectives: Thorium conjugates to target moieties such as antibodies are a promising new cancer therapy modality. SPECT imaging of Th-227 is desirable as an input to dosimetry, but is challenging because of the low yield of the emitted photons, complicated emission spectrum, small injected activities and its decay to Ra-223. The aim of this study was to develop a quantitative Th-227 SPECT reconstruction method to provide accurate estimates of absorbed doses in the different organs.

Methods: The reconstruction method is based on modeling image formation physics using the multiple-energy-range (MER) method originally developed for quantitative Y-90 bremsstrahlung imaging and recently applied to Ra-223. This MER modeling was incorporated into an OS-EM-based iterative reconstruction algorithm that models all the image degrading factors including attenuation, scatter and the collimator-detector response (CDR). We used the effective source scatter estimation method to model scatter and pre-computed CDR tables to model the interactions in the collimator-detector system including septal penetration and scatter. We evaluated the method using data simulated with the SIMIND Monte Carlo (MC) simulation program; the simulations modeled a Siemens Symbia dual-head SPECT system with a 9.5 mm thick NaI(Tl) crystal and a medium energy-low penetration (MELP) collimator. We simulated 6 spheres, of various sizes ranging from 1 to 28 cm³ and an activity concentration in each sphere of 2.5 kBq/cm³. The spheres were placed in a cylindrical phantom filled with either air or water. An acquisition energy window of 200-350keV, 64 equispaced angles over 360°, and total acquisition time of 32 minutes were simulated.

Results: We observed very good agreement between the MC simulated Th-227 projections and those obtained using the MER-based method. We evaluated the quantitative accuracy of the estimated activities in the different spheres and the different imaging scenarios. The percent errors of the reconstructed activities in the largest sphere were -1.5% and -1.88%, and -2.6% and -6.3% for the second largest sphere, when imaged in air and water, respectively. For the smallest sphere, the errors were -38% and -26% for the two cases, resulting largely from partial volume effects.

Conclusion: Th-227 imaging in the absence of its Ra-223 daughter is feasible when using good models of the image formation process. For better quantitative results, iterative based reconstruction methods with many updates are needed. Optimal image acquisition parameters such as the acquisition energy window width and center would likely improve quantification accuracy. When combined with previously-developed Ra-223 imaging methods and dual isotope reconstruction methods, this provides a platform for clinical quantitative Th-227 imaging. Research Support: N/A