Exhibit Hall | Forum 5
Purpose: The quantitative accuracy of yttrium-90 (⁹⁰Y) imaging and dosimetry is essential for developing clinical dose endpoints. Such studies require comprehensive investigations including quantitative reconstructions, empirical phantom studies, and validated Monte Carlo simulations. Thus, many studies have investigated the two fields separately. Our study presents a complete dose evaluation for a state-of-the-art ⁹⁰Y SPECT/CT methodology by consolidating both imaging and dosimetry methods.
Methods: A previously published empirical and pre-imaging protocol was completed. Subsequently, the background compartment and four largest spheres of a NEMA phantom were infused with water and 1.91 GBq of total ⁹⁰Y activity, then imaged with SPECT/CT. MLEM and OSEM reconstructions with attenuation, scatter, and collimator corrections were performed on MIMv7.1. Activity was self-calibrated to the phantom body. Optimal iterations were assessed with recovery coefficients (RC) on MLEM (1subset128iterations) whereafter OSEM reconstructions with varying equivalent iterations (8subset16iterations, 16subsets8iterations, 64subsets2iterations, 128subsets1iterations) were performed. Absorbed dose was calculated with the local deposition method (LDM) for every sphere on each reconstructed image. A validated Monte Carlo software reDoseMC created a water voxel S-value (VSV). Nominal density of 1.005 g/cm³ was used. A simulated NEMA phantom was convolved with the VSV to create reference doses. Dmean, D₇₀, D₅₀, and D₃₀ were compared between reconstructed images and reference doses.
Results: RCs ranged from 60.6-63.0% between all spheres at an optimal 128 iterations. The VSV had mean uncertainties of 1.47% at radial distances of 13.3 mm from the source voxel. LDM differences had maximum errors of 785%, 329%, 105%, and 58.8% for D₇₀, D₅₀, and D₃₀, and Dmean between five reconstructions.
Conclusion: Image-based, ⁹⁰Y SPECT/CT dosimetry was directly affected by imaging parameters. Large dose underestimations existed for all sphere inserts even in an optimized imaging protocol. The most consistent and accurate dose metric between reconstructions was Dmean while D₇₀ was the most variable.