Three-dimensional echocardiographic determination of left ventricular volumes and function by multiplane transesophageal transducer: dynamic in vitro validation and in vivo comparison with angiography and thermodilution.

The goal of this study was to validate 3-dimensional echocardiography by multiplane transesophageal transducer for the determination of left ventricular volumes and ejection fraction in an in vitro experiment and to compare the method in vivo with biplane angiography and the continuous thermodilution method. In the dynamic in vitro experiment, we scanned rubber balloons in a water tank by using a pulsatile flow model. Twenty-nine measurements of volumes and ejection fractions were performed at increasing heart rates. Three-dimensional echocardiography showed a very high accuracy for volume measurements and ejection fraction calculation (correlation coefficient, standard error of estimate, and mean difference for end-diastolic volume 0.998, 2.3 mL, and 0.1 mL; for end-systolic volume 0.996, 2.7 mL, and 0.5 mL; and for ejection fraction 0.995, 1.0%, and -0.4%, respectively). However, with increasing heart rate there was progressive underestimation of ejection fraction calculation (percent error for heart rate below and above 100 bpm 0.59% and -8.6%, P < .001). In the in vivo study, left ventricular volumes and ejection fraction of 24 patients with symmetric and distorted left ventricular shape were compared with angiography results. There was good agreement for the subgroup of patients with normal left ventricular shape (mean difference +/-95% confidence interval for end-diastolic volume 5.2+/-6.7 mL, P < .05; for end-systolic volume -0.5+/-8.4 mL, P = not significant; for ejection fraction 2.4%+/-7.2%, P = not significant) and significantly more variability in the patients with left ventricular aneurysms (end-diastolic volume 23.1+/-56.4 mL, P < .01; end-systolic volume 5.6+/-41.0 mL, P = not significant; ejection fraction 4.9%+/-16.0%, P < .05). Additionally, in 20 critically ill, ventilated patients, stroke volume and cardiac output measurements were compared with measurement from continuous thermodilution. Stroke volume as well as cardiac output correlated well to thermodilution (r = 0.89 and 0.84, respectively, P < .001), although both parameters were significantly underestimated by 3-dimensional echocardiography (mean difference +/-95% confidence interval = -6.4+/-16.0 mL and -0.6+/-1.6 L/min, respectively, P < .005).