Cross-modality validation of quantitative ⁹⁰Y imaging using a custom hepatopulmonary shunt phantom and patient data

In this study the discrepancies in dose estimations from a ⁹⁰Y-filled hepatopulmonary shunt phantom (HPSP) were analyzed using SPECT/CT, PET/CT, and PET/MR modalities. The dose variability was also evaluated among randomly selected individuals undergoing ⁹⁰Y-microsphere therapy. The HPSP comprised a right lung of 1,101 cm³, a left lung of 863 cm³, a healthy liver parenchyma of 690 cm³, and two intrahepatic tumours of 202 cm³ and 9.5 cm³. ⁹⁰Y-chloride activity was diluted in water and injected into the phantom. SPECT/CT, PET/CT, and PET/MR were acquired with the phantom immersed in water. Absorbed dose was calculated for phantom compartments and for patients with liver tumours who were treated with ⁹⁰Y-glass microspheres. The calculated lung shunt fraction (LSF) (12.0%) from the PET/MR images was closest to the actual LSF (11.03%), followed by PET/CT with 12.2% and SPECT/CT with 13.4% LSF. The deviation between the actual dose estimates and ⁹⁰Y-PET/MR was - 0.1% for the large tumour (202 cm³), -10.5% for the small tumour (9.5 cm³), -7.5% for the healthy liver parenchyma, and 6.1% for the lung. For ⁹⁰Y-PET/CT, the deviation was 4.8% for the large tumour, -7.5% for the small tumour, -15.5% for the healthy liver parenchyma, and 7.6% for the lung. In contrast, a significant disparity was observed in the ⁹⁰Y-SPECT/CT modality, with deviations of -6.3% for the large tumour, -65.3% for the small tumour, -10.7% for the healthy liver parenchyma, and 18.2% for the lung. In the patient study, the doses from ^99mTc-MAA SPECT/CT and ⁹⁰Y-PET/MR images were comparable, whereas ⁹⁰Y-SPECT/CT yielded significantly lower dose estimates. The findings obtained demonstrate that ⁹⁰Y-SPECT/CT generates significant inaccuracies in absorbed dose estimates mainly for small tumours. All modalities overestimated LSF and lung dose by less than 10% except SPECT/CT (18%). It is concluded that implementing an optimal imaging technique with robust scatter correction would mitigate the dose underestimate. ⁹⁰Y-PET/CT and ⁹⁰Y-PET/MR are the preferred methods for reliable dosimetry and precise dose-response assessment, especially owing to their robust correlation with ^99mTc-MAA SPECT/CT simulation.