Rellability of enhanced gated SPECT in assessing wall motion of severely hypoperfused myocardium: Echocardiographic validation

• Published in: Journal of nuclear cardiology 1998-07, Vol.5 (4), p.387
• Main Authors: Nichols, Kenneth, Depuey, E Gordon, Krasnow, Norman, Lefkowitz, David, Rozanski, Alan
• Format: Article
• Language: English
• Subjects: Regression analysis
• ISSN: 1071-3581; 1532-6551
• DOI: 10.1016/S1071-3581(98)90144-1

A method has been described for improving myocardial visibility on ^sup 99m^Tc-labeled sestamibi gated tomograms, even in the presence of severe hypoperfusion. It is essential to verify that images transformed in this manner truly depict the myocardium and do not contain image artifacts. This is especially important if transformed images are to be used to aid in the discernment of regional wall-motion abnormalities. All radially detected maximum counts were mapped automatically to the same brightness level for each cinematic frame. This produced tomographic cine images strongly suggestive of myocardium that appeared to translate but not to brighten from diastole to systole. Transformed scintigrams were compared with echocardiographic cine images of horizontal long axis and short axis views for 40 patients. Echocardiograms were of sufficient quality to allow comparison of radial distances from left ventricular center to midmyocardium for 15 short axis images and 25 horizontal long axis images. Readings were graded independently for 10 territories on a five-point scale (normal, mild-to-moderate hypokinesis, severe hypokinesis, akinesis, dyskinesis) of regional wall motion of original and enhanced scintigrams and echocardiograms. Comparison of echocardiographic and single photon emission computed tomographic (SPECT) locations of midmyocardial horizontal long axis points yielded a root-mean-square error value of 1.5±0.6 pixels (average absolute error, 11%±5%). SPECT versus echocardiographic wall-motion readings were compared by means of contingency table analysis. The loglikelihood ratio (G^sup 2^) was 109.3 (n=364; df=16) with probability of no association