PET optimization for improved assessment and accurate quantification of 90Y-microsphere biodistribution after radioembolization




90Y-microspheres are widely used for the radioembolization of metastatic liver cancer or hepatocellular carcinoma and there is a growing interest for imaging 90Y-microspheres with PET. The aim of this study is to evaluate the performance of a current generation PET/CT scanner for 90Y imaging and to optimize the PET protocol to improve the assessment and the quantification of 90Y-microsphere biodistribution after radioembolization.


Data were acquired on a Biograph mCT-TrueV scanner with time of flight (TOF) and point spread function (PSF) modeling. Spatial resolution was measured with a90Y point source. Sensitivity was evaluated using the NEMA 70 cm line source filled with 90Y. To evaluate the count rate performance, 90Y vials with activity ranging from 3.64 to 0.035 GBq were measured in the center of the field of view (CFOV). The energy spectrum was evaluated. Image quality with different reconstructions was studied using the Jaszczak phantom containing six hollow spheres (diameters: 31.3, 28.1, 21.8, 16.1, 13.3, and 10.5 mm), filled with a 207 kBq/ml 90Y concentration and a 5:1 sphere-to-background ratio. Acquisition time was adjusted to simulate the quality of a realistic clinical PET acquisition of a patient treated with SIR-Spheres®. The developed methodology was applied to ten patients after SIR-Spheres® treatment acquiring a 10 min per bed PET.


The energy spectrum showed the90Y bremsstrahlung radiation. The 90Y transverse resolution, with filtered backprojection reconstruction, was 4.5 mm in the CFOV and degraded to 5.0 mm at 10 cm off-axis. 90Y absolute sensitivity was 0.40 kcps/MBq in the center of the field of view. Tendency of true and random rates as a function of the 90Y activity could be accurately described using linear and quadratic models, respectively. Phantom studies demonstrated that, due to low count statistics in 90Y PET acquisition, the optimal parameters for the standard OSEM+PSF reconstruction were only one iteration and a postreconstruction filter of 6 mm FWHM, for both TOF and non-TOF reconstructions. Moreover, when TOF is included, the signal to noise ratio increased and visibility achieved 100% by the experienced observers and 93.3% according to the Rose model of statistical detection. In 50% of patients, TOF allowed the visualization of 90Y radioembolized lesions not seen without TOF, confirming phantom results. Liver activity was accurately quantified, with no significant differences between reconstructed and actual delivered activity to the whole-liver [mean relative difference (10.2 ± 14.7)%].


Qualitative and quantitative90Y PET imaging improved with the introduction of TOF in a PET/CT scanner, thereby allowing the visualization of microsphere deposition in lesions not visible in non-TOF images. This technique accurately quantifies the total activity delivered to the liver during radioembolization with 90Y-microspheres and allows dose estimation.