Background: Prolonged 3-dimensional echocardiography (3DE) acquisition time currently limits its routine use for calculating left ventricular volume (LVV) and ejection fraction (EF). Our goal was to reduce the acquisition time by defining the largest rotational acquisition interval that still allows 3DE reconstruction for accurate and reproducible LVV and EF calculation.
Methods: Twenty-one subjects underwent magnetic resonance imaging and precordial 3DE with 2 degrees acquisition intervals. Images were processed to result in data sets containing images at 2 degrees, 4 degrees, 8 degrees, 16 degrees, 32 degrees, and 64 degrees intervals by excluding images in between. With use of the paraplane feature, 8 equidistant short-axis slices were generated from each data set. The suitability of these short-axis slices for manual endocardial tracing was scored visually by 4 independent experienced observers. The LVV and EF were calculated by using Simpson's rule from 3DE data sets with 2 degrees, 8 degrees, and 16 degrees intervals, and the results were compared with values obtained from magnetic resonance imaging. The probability of 3DE to detect LVV and EF differences was calculated.
Results: All patients were in sinus rhythm with a mean heart rate of 72 bpm (SD + or - 12). The LV short-axis images obtained with 16 degrees rotational scanning intervals allowed LV endocardial tracing in all subjects. Good correlation, close limits of agreement, and nonsignificant differences were found between values of LVV and EF calculated with 3DE at 2 degrees, 8 degrees, and 16 degrees rotational intervals and those obtained with magnetic resonance imaging. At steps of 16 degrees, 3DE had excellent correlation (r = 98, 99, and 99), close limits of agreement (+ or - 38, + or - 28.6, and + or - 4.8), and nonsignificant differences (P =.5,.8, and.2) with values obtained from magnetic resonance imaging for calculating end-diastolic LVV, end-systolic LVV, and EF, respectively. Three-dimensional echocardiography with use of 16 degrees rotational intervals could detect 15-mL differences in end-diastolic volume with a probability of 95%, 11-mL differences in end-systolic volume with a probability of 92%, and 0.02 differences in EF with a probability of 95%.
Conclusions: The 3DE data sets reconstructed with images selected at 16 degrees intervals from data sets obtained at 2 degrees precordial rotational acquisition intervals allowed the generation of LV short-axis images with adequate quality for endocardial border tracing. Therefore precordial acquisition at 16 degrees intervals would be sufficient for the reconstruction of 3DE data sets for LV function measurement. This would reduce the acquisition time while maintaining enough accuracy for clinical decision making and would thus make 3DE more practical as a routine method.