[Variability and reproducibility in the determination of left-ventricular heart volume by cine-MR. A comparison of measurement methods].

Sixteen normal individuals were examined on a 1.5 T instrument using gradient echo sequences (cine-MR). The results of various computations were tested with respect to agreement and reproducibility of the methods. The ejection fractions calculated from bi-plane and 3-dimensional reconstructed volumes showed very low standard deviations of mean values (SD less than 2%). Volumes calculated from a single plane produced the greatest scatter of mean values (SC approximately 5%). In all sixteen subjects, the MR findings showed good reproducibility. There was very little inter-observer variability (variability less than 5%). A comparison of various methods showed that bi-plane and monoplane volume calculations obtained by MR in the cardiac axis are more accurate than those obtained by echocardiography (MR:R greater than 0.75, p less than 0.001; echo:R less than 0.75, p greater than 0.05).