Reference techniques for left ventricular volume measurement by three- dimensional echocardiography: Determination of precision, accuracy, and feasibility

The use of multiple in vitro reference methods to validate three- dimensional (3-D) echocardiographic techniques makes comparison difficult. In an attempt to establish a reference standard, we studied precision, accuracy, and feasibility of a true left ventricular (LV) volume measurement in six dog heart specimens using three techniques, called fluid, sheath, and cast. LV volumes ranged from 30 to 105 mL. Intraobserver variability was minimal in all combinations (1.26% to 2.8%) with a statistically insignificant tendency to higher values in the cast method. The cast method, however, exhibited significantly higher interobserver variability (5.78%) as compared to that ranging from 1.47% to 1.59% in the remaining two techniques. Regression analysis demonstrated high correlations among the three techniques assessed by 95% confidence limits and correlation coefficient (R² > 0.98, P < 0.01). Mean differences among the techniques (0.12 to 1.08 mL) were not significant. The fluid technique was easy to perform. The sheath technique required some practice. The cast method was sensitive to accurate preparation of a gelatin mixture. We conclude that the fluid and sheath techniques are precise, accurate, and feasible. We recommend their use as reference standards in laboratory LV volume measurement. Validation 3-D echocardiographic studies using either of these two techniques will be comparable. Although the accuracy of the cast technique is excellent, its lower precision makes it a second choice. It could be used in cases where an LV cavity cast is required and higher interobserver variability is acceptable.