Quantitative gated blood pool tomographic assessment of regional ejection fraction: definition of normal limits

Our aim was to select a method of analysis for gated blood pool tomography that reduced variability in a group of normal subjects, allowed comparison with normal limit files and displayed results in the bull's-eye format.

Abnormalities in left ventricular function may not be accurately detected by measures of global function because hyperkinesia in normal regions may compensate for abnormal regional function. Gated blood pool tomography acquires three-dimensional data and offers advantages over other noninvasive modalities for quantitative assessment of global and regional function.

Alternative methods for selecting the ventricular axis, calculating regional ejection fraction and choosing the number of ventricular divisions were studied in 15 normal volunteers to select the combination of parameters that produced the lowest variability in quantitative regional ejection fraction. Methods for quantitative comparison of regional ejection fraction with normal limit files and for display in the bull's-eye format were also examined.

A fixed axis (the geometric center of the ventricle defined for end-diastole and used for end-systole) gave ejection fractions that were significantly higher in the lateral wall versus in the septum, 82 +/- 8 (mean +/- 1 SD) versus 39 +/- 17 (p less than 0.001) at the midcavity and 66 +/- 11 versus 21 +/- 20 (p less than 0.001) at the base. A floating axis system (axis defined individually for end-diastole and end-systole and realigned at the center) gave more uniform regional ejection fraction: 63 +/- 6 versus 64 +/- 8 (p = NS) at the midcavity and 44 +/- 16 versus 45 +/- 15 (p = NS) at the base. The coefficient of variability for regional ejection fraction was consistently lower using a floating axis. Calculating regional ejection fraction by dividing the regional stroke volume by the end-diastolic volume of the region gave a lower coefficient of variability and a more easily understood value than dividing the regional stroke volume by the total end-diastolic volume of the ventricle. Although the variability was lower using five versus nine ventricular divisions, nine regions offer greater spatial resolution. Comparison of regional ejection fraction with normal data identified regions greater than 2.5 SD below the mean as abnormal. We described the two-dimensional bull's-eye format as a method for displaying the regional three-dimensional data and illustrated abnormalities in patients with prior myocardial infarction.

Gated blood pool tomography performed using a floating axis system, regional stroke volume calculation of ejection fraction and nine regions uses all the three-dimensional blood pool data to calculate regional ejection fraction, allow quantitative comparison with normal limit files, display the functional data in the two-dimensional bull's-eye format and demonstrate abnormalities in patients with myocardial infarction.