Evaluation of manufacture specific reconstruction algorithms available PET/CT imaging of Y-90 glass microspheres.

Objectives: PET/CT has been utilized to confirm the localization of Y-90 glass microspheres (YAZ) used for radioembolization. Different PET/CT manufactures utilize different reconstruction algorithms. The objective of this project was to evaluate one manufacture’s available PET/CT reconstruction algorithms for Y-90 YAZ to determine the optimal visual localization and concentration evaluation using the sealed vials provided rather than a fillable phantom.

Methods: YAZ were provided as TheraSphere^TM , which are in sealed glass vials surrounded by acrylic. Three vial sizes designated by the nominal activity of 20, 13, and 5 GBq were imaged as one bed acquisition for 20 minutes. The 20 and 5 GBq vials were imaged in the vertical position to ensure that all the YAZ were located in the bottom of the conical vial. The 13 GBq was evaluated in the horizontal position to test the accuracy of the reconstructions without partial volume correction. The Siemens Syngo reconstruction algorithm used was TrueX TOF using a matrix 440 x 440. This algorithm includes the following corrections: normalization, attenuation correction, decay, relative scatter correction, volume scaling, and resolution recovery. Iterative reconstructions of 2, 3, 4, 5 and 8 were performed with five subsets. To determine the PET/CT mean concentration, a threshold method was used in which the volume of interest was adjusted to match the expected volume. The mean concentration for each reconstruction performed was compared to the expected concentration as percent differences.

Results: The 4 iterative reconstructions generated for all three activities were the best visually. For both the 20 and 13 GBq vials, the mean concentrations generated from the reconstructions varied from the expected concentrations by 33.7%-39.3% and 34.9%-53.5% respectively. The mean concentration for the 5 GBq vial produced unexplained results, further assessment is warranted.

Conclusions: The results indicated a large discrepancy between the mean concentrations versus the expected concentrations for all three vials. More phantom testing would be required to determine if the scaling factor is the source of this discrepancy. Until such time, it is not recommended to use this method for validation.