Background: We evaluated the accuracy of planar radionuclide angiography and different count-based and space-based electrocardiogram (ECG)-gated blood-pool single-photon emission computed tomography (GBPS) algorithms for assessment of left ventricular end-diastolic volume (LVEDV), end-systolic volume (LVESV), and ejection fraction (LVEF) compared with the gold standard of cardiac magnetic resonance imaging (cMRI). The goal is to assess the accuracy of a recently developed GBPS algorithm.
Methods and results: Subjects had planar, GBPS, and cMRI sequentially. Datasets were processed by QBS software (Cedar-Sinai) and by MHI software (Montreal Heart Institute). Space-based approaches were used to compute LVEDV, LVESV, and LVEF. Count-based techniques were also used to assess LVEF. All results were compared to cMRI. Fifty-five patients (85% male; mean age 63 +/- 9 years) completed the study. LVEFs and their correlations to cMRI values were 43 +/- 12% (r = .82), 39 +/- 14% (r = .82), and 39 +/- 13% for MHI(space), QBS(space), and cMRI methodologies, respectively. LVEF by count-based methods also demonstrated good correlation to LVEF provided by cMRI (42 +/- 13%, r = .88 for MHI(count) and 46 +/- 15%, r = .84 for QBS(count)). Strong correlations were obtained for LVEDV (r = .96 for MHI and r = .92 for QBS) and for LVESV (.97 for MHI and r = .94 for QBS).
Conclusions: All Gated blood-pool SPECT algorithms had significant variation in estimating LVEF. Nevertheless our software provides good estimates of LV volumes and LVEF. Such software may, therefore, be applied to assess LV morphology and function.