Assessing dose variability in PET/CT-based voxel dosimetry for Y-90 radioembolization

This study aimed to evaluate the impact of calculation parameters on organ dose coefficients in image-based dosimetry for Yttrium-90 (Y-90) transarterial radioembolization (TARE). Patient-specific dose coefficient maps were generated using PET/CT images from six hepatocellular carcinoma patients using TOPAS Monte Carlo simulations. To reduce spurious activity, PET values below varying thresholds (0, 50, and 100 kBq/mL) were excluded, and the effect of these thresholds on dose coefficients was assessed. The influence of CT Hounsfield unit conversion was examined by varying the number of materials from 1 to 24. The effects of dose calculation method, physics models, and voxel sizes on dose coefficients were systematically investigated. Spurious activity was the dominant factor influencing the variability of the dose coefficient. Adapting the activity threshold resulted in the mean difference of 38.8 (15.2 %), 3.3 (11.9 %), and 0.1 (6.5 %) Gy/GBq for the tumor, liver, and lungs. The mean difference of lung dose coefficient (1.1 Gy/GBq, 67.2 %) between single and 24 materials. The dose calculation method, physics model, and PET voxel size had a relatively small impact on variability. This study systematically evaluated potential factors contributing variability of dose coefficient after Y-90 TARE, with spurious activity identified as a major factor.