Optimized SPECT Imaging of 224Ra α-Particle Therapy by 212Pb Photon Emissions

Abstract

In preparation for an α-particle therapy trial using 1-7 MBq of ²²⁴Ra, the feasibility of tomographic SPECT/CT imaging was of interest. The nuclide decays in 6 steps to stable ²⁰⁸Pb, with ²¹²Pb as the principle photon-emitting nuclide. ²¹²Bi and ²⁰⁸Tl emit high-energy photons up to 2,615 keV. A phantom study was conducted to determine the optimal acquisition and reconstruction protocol. Methods: The spheres of a body phantom were filled with a ²²⁴Ra-RaCl₂ solution, and the background compartment was filled with water. Images were acquired on a SPECT/CT system. In addition, 30-min scans were acquired for 80- and 240-keV emissions, using triple-energy windows, with both medium-energy and high-energy collimators. Images were acquired at 90-95 and 29-30 kBq/mL, plus an explorative 3-min acquisition at 20 kBq/mL (using only the optimal protocol). Reconstructions were performed with attenuation correction only, attenuation plus scatter correction, 3 levels of postfiltering, and 24 levels of iterative updates. Acquisitions and reconstructions were compared using the maximum value and signal-to-scatter peak ratio for each sphere. Monte Carlo simulations were performed to examine the contributions of key emissions. Results: Secondary photons of the 2,615-keV ²⁰⁸Tl emission produced in the collimators make up most of the acquired energy spectrum, as revealed by Monte Carlo simulations, with only a small fraction (3%-6%) of photons in each window providing useful information for imaging. Still, decent image quality is possible at 30 kBq/mL, and nuclide concentrations are imageable down to approximately 2-5 kBq/mL. The overall best results were obtained with the 240-keV window, medium-energy collimator, attenuation and scatter correction, 30 iterations and 2 subsets, and a 12-mm gaussian postprocessing filter. However, all combinations of the applied collimators and energy windows were capable of producing adequate results, even though some failed to reconstruct the 2 smallest spheres. Conclusion: SPECT/CT imaging of ²²⁴Ra in equilibrium with daughters is possible, with sufficient image quality to provide clinical utility for the current trial of intraperitoneally administrated activity. A systematic scheme for optimization was designed to select acquisition and reconstruction settings.

Keywords: Pb212; Ra224; SPECT; optimization; α-particle therapy.

Graphical abstract

FIGURE 1.

Captured (experiment) and Monte Carlo–simulated (cyan line) normalized energy spectra without (A) and with (B) ME collimator. Energy windows are marked with black lines on top axis. Stacked areas show contribution of imageable x-ray (70–90 keV) and γ-ray (239–241 keV) emissions, as well as scattered photons produced by 2,615-keV and 583-keV emissions. Presence of collimators greatly increased scatter; only low percentage of captured counts are from primary emissions. Supplemental Figure 1 shows for further panels.

FIGURE 2.

SSR as function of ordered-subset expectation maximization updates (logarithmic scale) for each phantom sphere for selected dataset (ME collimator, 240-keV window, and 12-mm gaussian filter), without (A) and with (B) SC. Dashed segments signify reconstructions where spheres did not show local intensity maximum. Setting that maximized total normalized SSR of all spheres was considered optimal (dashed vertical line). Supplemental Figures 4–7 show further panels.

FIGURE 3.

Example of image quality. (A–C) Transaxial slices at each exposure level: 90 kBq/mL (A), 30 kBq/mL (B), and 2 kBq/mL equivalent scan (C), captured using ME collimator and 240-keV window and reconstructed using SC and 12-mm gaussian filter at individually optimal number of iterations. (C) Spheres’ positions are rotated 60° clockwise.

FIGURE 4.

Maximum value in each sphere at each exposure level: 90 kBq/mL (A), 30 kBq/mL (B), and 2 kBq/mL (C) equivalent for selected dataset using ME collimator, 240-keV window, AC, SC, and 12-mm gaussian filter. Exposure-normalized count rates are shown on left axis, and absolute number of counts are shown on right axis. Dashed segments signify reconstructions where spheres did not show local intensity maximum. Supplemental Figures 11–14 show further panels. OSEM = ordered-subsets expectation maximization.

FIGURE 5.

Comparison of maximum voxel count rate in largest sphere at 90 kBq/mL (gray) and 30 kBq/mL (black). Difference in normalized count rates at 90 kBq/mL relative to 30 kBq/mL is shown on top of bars. Acquisitions were reconstructed using optimal settings at 30 kBq/mL for both concentrations.

FIGURE 6.

Comparison of maximized aggregated SSR values for each collimator and energy window combination, reconstructed without (A) and with (B) SC. Scatter-corrected images have higher SSRs and are plotted on wider scale. SCAC = Scatter and attenuation corrected.

FIGURE 7.

Effect of gaussian postprocessing filter widths on resulting SSRs, for example, setting with ME collimator and 240-keV window, each at individually optimal number of ordered-subset expectation maximization updates. Reconstruction without (A) and with (B) SC. Supplemental Figure 8 shows further panels. SCAC = Scatter and attenuation corrected.