To compare the quantitative algorithms Emory Cardiac Toolbox (ECTb), quantitative gated SPECT (QGS), layer of maximum counts (LMC), and left ventricular global thickening fraction (LVGTF) using gated myocardial tomography in the calculation of the left ventricular ejection fraction using the regression without truth (RWT) technique.
Materials and methods
Seventy-four consecutive patients were included in the study (59 males). All patients underwent stress–rest myocardial perfusion SPECT using 99mTc-tetrofosmin. Analysis of variance (ANOVA), the paired Student's t-test, the Pearson correlation coefficient and Bland–Altman were used for comparing the methods. The relative accuracy was performed by RWT.
ANOVA revealed a significant difference among the methods in calculating the ejection fraction. RWT showed that ECTb and QGS outperformed the other two methods. The ECTb was slightly better than QGS, and LMC was slightly better than LVGTF. QGS and ECTb achieved good correlations in end diastolic volume, end systolic volume and ejection fraction measurements. One-way ANOVA demonstrated that QGS was the only software program affected by the category of the perfusion summed stress score (SSS), P=0.038. The ejection fraction determined by the QGS, ECTb and LVGTF methods correlated significantly with defect size (r=0.545, P<0.0001; r=0.530, P<0.0001; and r=0.419, P<0.0001, respectively), but the LMC method was not significantly correlated (r=0.216, P=0.067).
There was a considerable variation among the quantitative gated SPECT methods in the evaluation of the ejection fraction. RWT revealed that the ECTb and QGS outperformed the other two methods with respect to the bias and precision of the measurements. Pair-wise correlations of the four methods ranged from mild to good with large agreement limits. Results of RWT provided important information in ranking the quantitative gated SPECT methods.