Standardising Y-90 PET quantification for a range of PET scanners: Results from the QUEST site assessment phase

589 Objectives The QUEST (Quantitative Uptake Evaluation in SIR-Spheres Therapy) study is a project aimed at determining the dose-response relationship in Y-90 radioembolization treatment of liver cancer. The initial stage consisted of establishing the viability of quantification of Y-90 PET imaging using a clinically relevant activity. Methods A NEMA IEC body phantom containing an 8:1 sphere-to-background concentration ratio of Y-90 was imaged on day 0 (~81 mCi), 3, 5 and 7. Reconstructions were conducted at each site and data analysed at the core laboratory in Sydney in an identical manner using in house software written in IDL. Quantitative measures, including total activity and concentration in hot spheres and background, were compared. Assessments were made based on brand and generation of machine, and software functions for reconstruction. Results 50 institutions from 13 nations provided data from different generation machines (time of flight (ToF) and non-ToF) from each of the three vendors (GE, Philips and Siemens). Analysis showed TOF scanners, irrespective of brand, provided comparable and superior quantification. Non-TOF machines using 3D iterative reconstruction and resolution recovery (RR) were slightly inferior. Significant deterioration in recovery and quantification was found on non-TOF scanners employing 2D iterative reconstruction algorithms. Treatment of randoms correction appears to have a large impact on quantification as seen with certain editions of software. Background concentration appears well estimated to levels up to at least 2.7 µCi/ml. Conclusions Y-90 PET appears quantitative for measures of uniform concentrations over a large range on current generation scanners from all vendors, as well as non-ToF scanners employing 3D iterative reconstruction with RR. All scanners appear to underestimate concentrations in spheres of up to 37 mm diameter which may be due to issues of random coincidence correction and image bias or convergence.