RUBY-10: A cross-comparison study of 10 software tools for quantification of Rb-82 myocardial perfusion PET

1693 Objectives To cross-compare MBF estimates from Rb-82 PET data for 8 models implemented in 10 software tools (SWT): Carimas, Corridor4DM, FlowQuant, HOQUTO, ImagenQ, MunichHeart, PMOD, QPET, syngo.MBF, UW-QPP. Methods Rest and stress Rb-82 PET/CT scans (10MBq/kg, 3D acquisition) of 48 patients with suspected or known CAD were analyzed in 10 centers. Each center used one of 10 SWT to analyze global and regional (LAD, LCx and RCA) MBF. Values were considered to agree if they simultaneously had an intraclass correlation coefficient (ICC)>0.75 and a difference Results MBF estimates from all 7 SWT implementing 1TCM as in Lortie et al. (2007) were in agreement with one another on studied levels. Results of the model by El Fakhri et al. (2009) as in Corridor4DM were also in excellent agreement with the Lortie’s 1TCM, so were the results of the axially-distributed blood flow model (Alessio et al. 2011) in UW-QPP, yet, with occasional differences of >20% of median values from the Lortie’s 1TCM. The MBF estimates from 2TCM (PMOD, Herrero et al. 1992), and net retention models (ImagenQ - Yoshida et al. 1996, MunichHeart - Lautamaki et al. 2008), though producing similar results were less correlated with Lortie’s 1TCM (ICC Conclusions SWT using the same 1TCM (Lortie et al. 2007), provide nearly equivalent results in measuring global and regional MBF (rest and stress), suggesting that they may be clinically interchanged for the benefit of multicenter trials in future.