TO THE EDITOR:
In a recent article (1), my friend and colleague Grant Gullberg presented a method for the calculation of ejection fraction in gated myocardial perfusion SPECT studies. The crucial point is the avoidance of edge detection. In their method, the authors used the maximal activity plane as the ventricular delineator. The originality of the method resides in the latter, not in the former, since avoidance of edge detection for ventricular delimitation has been described before. Several studies (2–5) have used a method in which the delineator was the first moment of the count density distribution across the myocardial wall. I will admit that the Japanese reference could easily have been missed.
REPLY:
We truly regret having overlooked the work of Michael Goris. As is mentioned in his letter, he also proposed a method for estimating left ventricular ejection fraction (LVEF) without having to measure the edge of the intraventricular cavity. Our work was aimed at improving calculation of the LVEF for small hearts by developing a method that was less sensitive to partial-volume effects, whereas his work was aimed at developing a method that would improve the calculation of the LVEF in the case of perfusion defects. Because both methods do not detect edges, they should both work better in the case of perfusion defects since no interpolation is required for missing data.
The Goris method (1) processes the 3-dimensional gated nuclear tomographic images to eliminate nonmyocardial structures and to locate the center of the left ventricular cavity in the end-systolic time bin. The position of the centroid of the counting rate distribution along all radii emanating from this center is estimated for each time bin. All of the centroid distances at end-systole and end-diastole are used to estimate the corresponding volumes for calculating the LVEF (2). On the other hand, our method transforms the 3-dimensional image in Cartesian coordinates into an image in prolate spheroid coordinates. In this coordinate system, the wall of the left ventricle appears as a plane. The maximum activity along each radius can then easily be determined. We present a mathematic derivation that shows that the position of this maximum value can be used to accurately estimate the LVEF for large and smaller hearts. As Michael Goris correctly points out, this is the innovation in our method.
LVEF calculated using the Goris method of calculation correlates well with LVEF calculated using conventional planar equilibrium radionuclide angiocardiography, even in the case of perfusion defects. Our method gave results that correlated well with the quantitative gated SPECT method (3,4) for large hearts and was shown to be more accurate for smaller hearts. However, our work was not evaluated for perfusion defects, but the method may do well for the same reason that the Goris method does well in the presence of severe perfusion defects (2).
We thank Michael Goris for bringing his excellent work to our attention. Both his method and our method have significant merit for improving the calculation of LVEF from nuclear scintigraphic images. Further work is needed to evaluate these techniques for different sizes of hearts with various abnormalities.