Self-calibration methods for voxel-based dosimetry using ⁹⁰Y Bremsstrahlung SPECT/CT imaging following selective internal radiation therapy (SIRT) of liver tumors

Objectives: Dosimetry using ⁹⁰Y Bremsstrahlung SPECT/CT imaging following SIRT of unresectable liver tumors can be useful in assessing the delivered dose distribution and quantifying metrics for tumor response and potential radiation induced effects to normal liver tissue and lungs. The aim of this study was to examine four calibration methods for converting the amount of counts in each voxel to activity for the dosimetric analysis.

Methods: Voxel-based dosimetry using commercial software (PLANET; DOSIsoft) was performed on ⁹⁰Y Bremsstrahlung SPECT/CT images of 20 patients who received SIRT for hepatic lesions with either resin or glass microspheres. The absorbed dose to the tumor (GTV) and normal liver were analyzed using a local deposition model (LDM) implementation with 4 different self-calibration methods defining the amount of activity in the SPECT volume of interest: entire imaged field of view (FOV), chest/abdomen, lung/liver, and total liver. Statistical analysis was performed using linear correlation and Bland-Altman plots to compare the 4 self-calibration methods using dose volume histogram (DVH) indices of mean, D50, D20 and D70 for the GTV and normal liver. Results: The mean dose to GTV and normal liver calculated using the chest/abdomen self-calibration method were within 10% of the SPECT FOV (ranges, 5-21% for GTV and 5-26% for normal liver). The mean dose calculations for GTV and normal liver using the liver/lung and liver-only self-calibration methods were within 9% of each other (ranges, 2 - 17% for both GTV and normal liver), while they differed substantially from the SPECT FOV method (t-test, P<0.05). There was strong linear correlation (R² > 0.99) for all DVH metrics in using the chest/abdomen versus SPECT FOV methods. The percentage bias in mean dose to GTV and normal liver calculated by chest/abdomen compared with the SPECT FOV had a mean value of 10% (95% limits of agreement between 2% and 18%). Conclusions: The mean GTV as well as mean normal liver tissue dose calculated using the SPECT FOV and the chest/abdomen self-calibration methods respectively yielded very similar results. The liver/lung and liver-only methods differed substantially from the other two methods while they were similar to each other. This study suggests that absorbed dose calculations can be significantly influenced by the choice of self-calibration method and therefore standardization and consistency are essential in Bremsstrahlung SPECT/CT imaging following ⁹⁰Y SIRT.

• Download figure
• Open in new tab
• Download powerpoint