Abstract
41
Objectives: Targeted thorium conjugates are currently being investigated as a new class of alpha-radiopharmaceuticals. The decay of Th-227 produces Ra-223, which in turn redistributes inside the body. In order to estimate the dose received by the various organs, accurate quantification of Th-227 and Ra-223 activities is essential. Simultaneous quantitative SPECT imaging of Th-227 and Ra-223 is challenging because of the low yield of photons, complicated emission spectra and the crosstalk from one radionuclide to the other. The aim of this study was to develop a dual isotope quantitative SPECT Th-227 and Ra-223 reconstruction method that accounts for the crosstalk from one radionuclide to the other. Materials and Methods: We used a realistic XCAT digital phantom. We separately simulated low noise Th-227 and Ra-223 projection data for heart, liver, lungs and spleen, using the SIMIND MC simulation program. Post-simulation scaling and summing was used to efficiently model activity of the two radionuclides at 4, 72, 339, 675 hours post injection. We modelled an injected Th-227 activity of 2.8 MBq, a Siemens Symbia camera with an MEGP collimator, acquisition at 64 views over 360° and an acquisition duration of 30 minutes per bed position. We used energy windows of 200-350 keV for Th-227 and 75.6-92.4 keV, 146.3-161.7 keV, 255.6-282.6 keV for Ra-223. We developed and implemented an iterative reconstruction method that compensated for Attenuation (A), Scatter (S), Collimator-Detector response (CDR), including the geometric and septal penetration and scatter, and a crosstalk estimation model from each radionuclide to the other. Projection images, acquired in the Th-227 and Ra-223 acquisition energy windows, were first reconstructed using the OS-EM algorithm with A, S and CDR compensations assuming that all photons were from Th-227 and Ra-223, respectively. A crosstalk estimate for each radionuclide was obtained by projecting these reconstructed images to the other radionuclide’s window. Since the reconstructed images were not the true activity distribution, we repeated the crosstalk estimation process several times by including the crosstalk estimate as an additive correction component to the reconstruction until the change in the estimate was small. The converged crosstalk estimates were then used as an additive correction in a final reconstruction that yielded the final Th-227 and Ra-223 activity images. Results and Discussion: We reconstructed the low noise data using the iterative method described above and evaluated the bias in the estimated Th-227 and Ra-223 activities in the different organs for the different acquisition time points. For organs with high activities > 10 kBq of Ra-223 and > 80 kBq of Th-227, the biases in both radionuclides’ estimates were less than 5%. We also calculated the mean bias and coefficient of variation (COV) over all noise realizations. Mean biases and COV of Ra-223 and Th-227 estimates were ~ 8%, 21%, and 10%, 30% respectively, averaged over all organs, time points and noise realizations.
Conclusions: We developed and evaluated a quantitative SPECT dual isotope reconstruction method for Th-227 and Ra-223 that accounts for crosstalk contamination. Organs that had activities greater than 10 kBq and 80 kBq of Ra-223 and Th-227, respectively, had a bias of ~5% and precision of ~20%. Compensation for partial volume effects would further improve quantification results of small organs. These results suggest that clinical simultaneous quantitative Th-227 and Ra-223 SPECT imaging is feasible.