Background: Cardiac gated blood pool single photon emission computed tomography (GBPS) better separates cardiac chambers compared with planar radionuclide ventriculography (PRNV). We have developed a completely automatic algorithm to measure quantitatively the left ventricular ejection fraction (LVEF) from gated technetium 99m-red blood cells (RBC) GBPS short-axis 3-dimensional image volumes.
Methods and results: The algorithm determines an ellipsoidal coordinate system for the left ventricle and then computes a static estimate of the endocardial surface by use of counts and count gradients. A dynamic surface representing the endocardium is computed for each interval of the cardiac cycle by use of additional information from the temporal Fourier transform of the image data sets. The algorithm then calculates the left ventricular volume for each interval and computes LVEF from the end-diastolic and end-systolic volumes. The algorithm was developed in a pilot group (N = 45) and validated in a second group (N = 89) of patients who underwent PRNV and 8-interval GBPS. Technically inadequate studies (N = 38) were rejected before grouping and processing. Automatic identification and contouring of the left ventricle was successful in 121/172 patients (70%) globally and in 76/89 patients (85 %) in the validation group. Correlation between LVEFs measured from GBPS and PRNV was high (y = 2.00 + 1.01x, r = 0.89), with GBPS LVEF significantly higher than PRNV LVEF (average difference = 2.8%, P < .004).
Conclusions: Our automatic algorithm agrees with conventional radionuclide measurements of LVEF and provides the basis for 3-dimensional analysis of wall motion.