Reproducibility of left ventricular volume determinations with use of a semi-automated system.

Reliability of a computer-assisted system for determination of left ventricular volumes was judged by multiple measurements of rotation ellipsoids, cadaver hearts, and cineangiograms from patients. The volume measurements in cadaver hearts provided a volume correction factor necessary for reproducible results. Variation coefficient for intraobserver and interobserver variability did not exceed 2.3% when calculated using rotation ellipsoids and was highest at 12.0% for the end-systolic volumes derived from patient films. When appropriate calibration methods are employed, different observers can make reliable left ventricular volume measurements aided by such systems.