Radionuclide left ventricular ejection fraction: a comparison of three methods.

Three commercially available computer programs (a semiautomatic method, a manual method, and a regional method) were used to calculate left ventricular ejection fraction from the equilibrium multiple gated radionuclide ventriculograms obtained from 24 normal male subjects and 20 men with heart failure. In the normal subjects the ejection fraction values calculated by each method were significantly different (mean SD) difference between semiautomatic and manual 3.3 (5.8); between semiautomatic and regional 12.0 (6.3); and between manual and regional 8.7 (6.9]. In the patients with heart failure the ejection fraction values calculated by the semiautomatic method differed significantly from those calculated by the manual and regional methods (mean (SD) difference between semiautomatic and manual 3.4 (4.7); between semiautomatic and regional 4.9 (4.9); and between manual and regional 1.5 (6.2]. The ejection fraction values obtained by the semiautomatic method were generally higher and more consistent than those derived from the manual and regional methods. An ejection fraction of greater than or equal to 50% with the semiautomatic method would be regarded as normal but if the same normal range was applied to the regional method nine (38%) of the 24 normal subjects would appear to have an abnormal left ventricular function. Clinicians should be aware that the method used to generate a time-activity curve is an important consideration in the calculation of ejection fraction. Each centre should establish its own range and reproducibility for the method it uses to measure ejection fraction. These values should not be assumed to apply to any other method.