First-pass radionuclide angiographic analysis with two regions of interest to improve left ventricular ejection fraction accuracy.

Abstract

First-pass radionuclide angiographic (FPRNA) analysis, using the standard, single-fixed region of interest (ROI) drawn at end-diastole, often underestimates the left ventricular ejection fraction (LVEF) as determined by other standard techniques. This study examined the hypothesis that correction for the anatomic motion of the aortic valve plane toward the apex during systole, which results in improper inclusion of aortic counts within the single-fixed ROI, using a two-ROI method to compensate for this motion would eliminate this underestimation. In 70 patients who underwent FPRNA and planar gated equilibrium radionuclide angiography (GERNA) on the same day, Fourier transform phase and amplitude images were used to generate functional maps of the aorta and the left ventricle on the FPRNA representative cycle. The region of low amplitude between the aorta and left ventricle, which corresponds to the degree of aortic valve plane motion, was used to guide the manual placement of two ROIs. The first was over the left ventricle at the end-diastole including the aortic valve plane area, and the second was a smaller end-systolic ROI drawn over the first ROI, excluding the valve plane area. Both the fixed- and dual-ROI FPRNA methods had excellent correlation with GERNA (r = 0.92 and 0.91, respectively). The mean FPRNA LVEF using a fixed ROI (45% +/- 14%) was significantly lower than GERNA (51% +/- 15%, p < 0.001), but the mean LVEF calculated from the dual-ROI (51% +/- 14%) was essentially identical to those obtained with GERNA. The method of manual placement of the two ROIs had extremely high levels of inter- and intraobserver reproducibility (r = 0.98 and 0.99, respectively). Despite good correlation, the standard, fixed-ROI method of FPRNA analysis systematically underestimates the LVEFs of GERNA. This problem can be eliminated by taking into account valve plane motion during the cardiac cycle by using Fourier-guided, dual-ROI analysis on FPRNA. These differences in methods and results should be considered when substituting or comparing LVEFs derived from these techniques.