Abstract
1727
Objectives The performances of QBS software (Cedars Sinai) for the quantification of cardiac function with Gated blood pool SPECT (GSPECT RNA) radionuclide angiography (RNA) are better with iterative reconstruction combined to resolution recovery (3D-Flash) as compared to filtered backprojection reconstruction (FBP). In this study, we aim to compare its performance when using iterative reconstruction (OSEM) without resolution recovery versus FBP for the measurement of left ventricular ejection fraction (LVEF).
Methods The study included 140 patient addressed for LV function evaluation with planar (left anterior oblique view, planarLAO) and GSPECT RNA. GSPECT RNA studies were reconstructed using both FBP and OSEM and then processed with QBS software. For each of the 2 reconstruction methods, the 3 different LVEF values provided by QBS were noted: LVEF_actratio (ratio of activities of maximal activity threshold - MAT method), LVEF_volratio (ratio of LV volumes of MAT method), and LVEF_Gd (gradient method). PlanarLAO LVEF was considered gold standard.
Results LVEF provided by planarLAO (58±10%) is significantly lower (P<0.01) than GSPECT RNA LVEF_actratio_OSEM (70±15%), LVEF_actratio_FBP (70±15%), LVEF_Gd_OSEM (63±14%), and LVEF_Gd_FBP (62±14%). But it is not different (P>0.05) from LVEF_volratio_OSEM (60±14) and LVEF_volratio_FBP (59±14%). PlanarLAO LVEF is highly correlated to both OSEM and FBP GSPECT RNA LVEF (P <0.0001): LVEF_actratio_OSEM (r=0.78), LVEF_volratio_OSEM (r=0.77), LVEF_Gd_OSEM (r=0.78), LVEF_actratio_FBP (r=0.74), LVEF_volratio_FBP (r=0.72), LVEF_Gd_FBP (r=0.73). Note all correlations are slightly higher with OSEM than with FBP. On Bland Altman analysis, the limits of agreement were slightly tighter with OSEM than with FBP versus planarLAO: -11.3±18.3% and -11.3±19.8% for LVEF_actratio , -1.5±17.1% and -0.4±19.1% for LVEF_volratio_FBP, and -4.5±17.5% and -3.1±19.4% for LVEF_Gd_FBP, respectively.
Conclusions QBS performances seem better when using OSEM than FBP reconstruction.