Abstract

Background

Gated myocardial SPECT is a well recognized tool for coronary disease evaluation. The dedicated softwares tend to present discrepancies in the measurement of left systolic functional parameters. Objective: to evaluate the performance of the Emory Toolbox® regression equations (ECTb) with respect to Cedars QGS® and echocardiographic values, as an independent reference.

Methods

228 myocardial SPECTs were analyzed, 40 with concomitant resting echocardiography (Simpson method), obtaining LVEF by QGS and ECTb version 4.0, applying regression equations R0 (original) and R1 (8 frames / min). Pearson or Spearman correlation was performed according to normality and Bland-Altman dispersion analysis.

Results

A: LVEF means: manual resting QGS were 50 ± 23%, R0 64 + 25% and R1 62 ± 22%; in stress 53 ± 21%, 70 ± 22% and 67 ± 18%, respectively. The automatic resting QGS were 55 ± 25%, R0 64 ± 25% and R1 58 ± 25%; in stress 59 ± 22%, 70 ± 22% and 63 ± 22%, respectively. B. Correlations: The LVEF echo vs manual SPECT had r of 0.574 for QGS (p <0.001), for R0 0.574 (p = 0.0002) and for R1 0.654 (p <0.0001). In automatic LVEF, the r was 0.751 for QGS vs R0 and 0.761 vs R1. In the manual LVEF, the r was 0.868 for QGS vs R0 and 0.877 vs R1. All the correlations had p <0.005 and the linear regression analysis did not show differences between slopes, but between intercepts (p <0.0001). C. Dispersion. The average difference between the manual LVEF QGS with both equations with R0 was 39.6% ± 28.2% and with R1 of 12.5% ± 46.7%; greater dispersion was observed at lower values.

Conclusion

In manual processing, the regression equation R1 ECTb fits better than QGS with echocardiography and can be used to obtain LVEF similar to those calculated by QGS, however, it presents differences greater than 10% in low values, so it does not seem recommended to follow-up of patients with various processing softwares, despite the use of regression.