Objective: Two different algorithms operating in three-dimensional space, one dependent on surface detection developed at Cedars-Sinai Medical Center (CS) and another dependent on statistical parameters and developed at Stanford University Medical School (SU), were compared in the same patients to assess the left ventricular volumes and the left ventricular ejection fractions (LVEFs) from gated single-photon emission tomography (SPECT) myocardial perfusion images.
Methods: Perfusion SPECT images gated in eight time bins were recorded in 40 patients with coronary artery disease 60 minutes after the injection of 925 MBq 99mTc-labeled tetrofosmin at rest. The LVEF values were validated against planar gated 99mTc-labeled blood pool studies (ERNA).
Results: The software success rates were 95% (38/40 patients) for CS and 100% for SU. Agreement between LVEFs measured with CS and SU and agreement between both methods and ERNA were excellent (LVEF[CS] = 0.89LVEF[SU] + 6.21, r = 0.93; LVEF[SU] = 0.92LVEF[ERNA] + 0.99, r = 0.94; and LVEF[CS] = 0.88LVEF[ERNA] + 4.58, r = 0.93). Bland-Altman plots showed that differences between LVEFs from SU and CS and from ERNA were similar across a wide range (20% to 80%) of LVEF values. No relationship between these differences and the severity of perfusion defects was observed. For left ventricular volumes, linear regression analysis showed an excellent correlation between both methods (end-diastolic volume r = 0.97 end-systolic volume r = 0.98), but systematically higher values were obtained with SU (p = 0.013).
Conclusion: Measurements of LVEF obtained with CS and SU correspond well with those from the standard, ERNA, even in patients with severe perfusion defects. A close relationship is observed between SU and CS when left ventricular volumes are considered. Measurements of LVEF (and left ventricular volumes) should be considered as an integral part of myocardial perfusion studies whenever possible.