Evaluation of ejection fraction measurements in gated cardiac imaging using dynamic cardiac phantoms

The purpose of this study was to evaluate the accuracy of left ventricular ejection fractions (LVEFs) obtained using gated planar imaging of dynamic cardiac phantoms and measured with a hybrid boundary detection technique. Water mixed with Tc-99m pertechnetate was used to fill the ventricles, and from 19 ml to 52 ml was added to the LV chambers at end diastole so that LVEFs of 24% to 56% were produced. Three different low energy collimators (high resolution, ultra-high resolution, and super high resolution parallel beam collimators) were evaluated. A Metz filter was used for spatial smoothing and a hybrid boundary detection algorithm was used for generating regions of interest (ROIs) in the twenty frames of images per R-R interval accumulated from 410 to 460 beats. The boundaries of the LV chambers were determined by combined first- and second-difference operators weighted by a hybrid enhancement weight α(0<α<1). The counts in the ROIs were used to calculate LVEFs. In general, the higher the value of a, the greater the deviations from actual EFs. The second-difference (α=0) method resulted in more accurate values than the first-difference (α=1) method. No large differences in the EF measurements were observed between the collimators. In the presence of background activity, the nearly optimal weight for three different collimators was α=0.5