Assessment of quantitative methods for 2-dimensional echocardiography.

Abstract

Several 2-dimensional echocardiographic (2-DE) methods were tested in vitro for accuracy of linear and cross-sectional measurements and in vivo for left ventricular (LV) volume reconstruction. With 2-DE instrument settings at low and high gains and with precise in vitro calibrations, we studied myocardial slice thickness (3.0 to 10.0 mm). The 2-DE myocardial thickness was measured by leading-trailing, trailing-leading, and leading-leading methods. Regression analysis of 2-DE versus direct measurements yielded excellent correlations for all 3 methods (r greater than 0.985), with interobserver variability less than 3%. Accuracy of measurement was satisfactory only for the leading-leading method (3 and 6% error at low and high gains, respectively); other methods substantially over- or underestimated thickness. Thin myocardial slices (less than 1 mm thick) were applied to cylinders and fixed in formalin to produce precise cavity areas (1.8 to 7.0 cm2). Regression analysis of 2-DE versus actual cavity area gave high correlations (r greater than 0.970), and low interobserver variability (less than 4%) for the inner edge and leading edge methods, but the leading edge method was the most accurate (1.3 to 2.5% error). In vivo LV volumes in 7 anesthetized dogs were compared with 2-DE and cineangiography. Good correlations (r = 0.92) were obtained, but the inner edge method underestimated angiographic volume, whereas the leading edge method reduced the magnitude of underestimation. Thus, the leading edge method for 2-DE is most accurate not only for linear and cross-sectional measurements of the myocardium, but also for application to in vivo LV volumes.