Influence of image quality on the accuracy of real time three-dimensional echocardiography to measure left ventricular volumes in unselected patients: a comparison with gated-SPECT imaging.

BACKGROUND: Patient selection, often restricted to those with ideal image quality, and timing of studies in relation to reference methods may limit clinical applicability of cardiac volumes derived from 3D echocardiography.

METHODS: To test the influence of image quality on LV volumes by real time 3DE (RT3DE), we compared results obtained by RT3DE to those from gated-SPECT imaging in 64 consecutive patients referred for clinically indicated nuclear perfusion imaging. To minimize hemodynamic effects, RT3DE was performed immediately following G-SPECT. LV volumes by RT3DE were calculated using at least three orthogonal plane pairs. Image quality was rated as good if 75-100% of the endocardial border was visualized, fair if 60-74% was visualized, and poor if 50-60% was visualized.

RESULTS: Image quality was good in 25 (39%), fair in 20 (31%), and poor in 13 (20%) patients. Six patients (9%) were excluded for uninterpretable echo images. For the entire cohort, EDV and ESV agreed closely (all P = NS). When stratified by image quality, the EDV and ESV of those with good and fair image quality agreed closely with minimal bias (average 1 +/- 9 mL and 2 +/- 7 mL, respectively). Poor image was associated with less strong agreement and much greater bias for EDV and ESV (7 +/- 25 mL and 7 +/- 20 mL, respectively).

CONCLUSIONS: When applied to patients studied in routine clinical practice, LV volumes by RT3DE compare favorably to G-SPECT. RT3DE results are more reliable when >60% of endocardium is visualized.