Variability in frame by frame analysis of left ventricular wall motion from contrast angiograms

Background. Measurement of the timing of left ventricular (LV) wall motion, of asynchrony, and of diastolic function from contrast angiograms requires delineation of the endocardial border frame by frame through the cardiac cycle. This study was performed to determine the magnitude of intraobserver and interobserver variability in manual border tracing, and to measure the impact of this variability on the derived functional parameters. Methods. The contrast ventriculograms of 25 patients with coronary artery disease (CAD) or with normal coronary arteries were analyzed frame by frame, by two observers or twice by the same observer. Motion was measured using the centerline method at each twelfth of systole and of diastole. Variability was calculated as the absolute difference between repeated measurements of: wall motion, asynchrony, and the time at which each region of the LV reached 10%, 50%, and 100% of peak contraction, and 50% of filling. Results. Intraobserver and interobserver variability in wall motion were similar, and varied with time in the cycle, and with location on the LV contour. Variability was highest at end systole, when it averaged 8% of the normal mean for wall motion. Variability in timing was highest at peak contraction; however, the variability in measuring asynchrony averaged only 18 msec. Conclusion. Analysis of the magnitude and synchrony of regional LV wall motion through the cardiac cycle from contrast ventriculograms can be performed with reproducibility comparable to that at end systole.