Reproducibility of left ventricular measurements with acoustic quantification: the influence of training

Acoustic quantification (AQ) of two-dimensional (2-D) echocardiograms provides online estimation of left ventricular (LV) size and function. However, edge detection with AQ is influenced by gain settings and is therefore operator dependent. Our purpose was to compare AQ and conventional 2-D echo measurements of LV size and function obtained by different operators and to evaluate the influence of training on these measurements. A cardiac sonographer without previous experience with the AQ system was trained by an experienced operator. Twenty-two normal males (age, 28 +/- 4 years) participated in the study. Images were recorded with conventional 2-D and AQ echo from the short-axis and apical four-chamber views. During the initial training period, five subjects were imaged by the sonographer under the supervision of the trainer. At the initial study session, 12 subjects were imaged independently by the two operators. Following a second training period with five different subjects, the same initial 12 subjects were again imaged at a second study session. LV cavity areas were traced from the conventional 2-D echocardiograms and measured from the AQ waveforms at end-diastole and end-systole. Volumes were calculated using the single-plane area-length method. Ejection fraction (EF) was calculated from volumes. Reproducibility was determined by comparing the variability of AQ and conventional 2-D echo measurements obtained at the two sessions. A second training session reduced the operator variability only of the short-axis end-diastolic area measurement (17 +/- 11% vs 6 +/- 5%, P < 0.025). We conclude that a single training session may be adequate for the reproducible estimation of ventricular volumes with the AQ method.