Determination of right ventricular ejection fraction from reprojected gated blood pool SPET: comparison with first-pass ventriculography

Abstract.

Gated blood pool (GBP) studies are widely available and relatively inexpensive. We have previously published a simple and convenient method for measuring left ventricle ejection fraction (EF) with increased accuracy from single-photon emission tomography (SPET) GBP scans. This paper describes an extension of this method by which right ventricular EF may also be measured. Gated SPET images of the blood pool are acquired and re-oriented in short-axis slices. Counts from the left ventricle are excluded from the short-axis slices, which are then reprojected to give horizontal long-axis images. Time-activity curves are generated from each pixel around the right ventricle, and an image is created with non-ventricular pixels "greyed out". This image is used as a guide in drawing regions of interest around the right ventricle on the end-diastolic and end-systolic long-axis images. In 28 patients, first-pass ventriculography studies were acquired followed by SPET GBP scans. The first-pass images were analysed a total of four times by two observers and the SPET images were analysed three times each by two observers. The agreement between the two techniques was good, with a correlation coefficient of 0.72 and a mean absolute difference between first-pass and reprojected SPET EFs of 4.8 EF units. Only four of the 28 patients had a difference of greater than 8 EF units. Variability was also excellent for SPET right ventricular EF values. Intra-observer variability was significantly lower for SPET than for first-pass EFs: standard error of the estimate (SEE)=5.1 and 7.3 EF units, respectively (P<0.05). Inter-observer variability was comparable in the two techniques (SEE=5.2 and 6.9 EF units for SPET and first-pass ventriculography, respectively).