Comparison of scintigraphic count-volume and geometric methods for measuring left ventricular ejection fraction.

Abstract

A comparison of geometric and count-volume scintigraphic methods for measuring left ventricular ejection fraction (EF) is presented. The particular count-volume method used is a modification of an established first-pass technique. It is shown that a modification of the technique is necessary to prevent a positive bias in the EF estimate due to Poisson noise. Using the same simulated data, EF estimates by the modified technique were always closer to the actual EF volume than were the estimates by the original method. Maximum Likelihood estimates were also more accurate than were the other two estimates. In 56 patients with either acute myocardial infarcts or unstable angina pectoris, the scintigraphic single-plane gated geometric method tended to overestimate EF in each of the patient groups when compared with the geometry-independent count-volume estimates. The results also point out that the error in the geometric estimates tended to be greater for larger ejection fractions and appears to be due to the systematic underestimation of end-systolic volumes.