Determination of ejection fraction from contrast echocardiography using videodensitometry in an in vitro model.

To examine whether ejection fraction can be estimated by videodensitometric analysis of contrast washout from a chamber, a model was constructed consisting of a 60-ml syringe and a modified Harvard respirator. Ejection fraction could be adjusted mechanically. Echocardiographic images were recorded on videotape and analyzed on an offline computerized analysis system. To calibrate the relationship between target concentration and videodensity on our sector scanner-videodensity analysis system, we used a phantom with constant attenuation and eight cones with known, varying relative scatterer concentration. The time-gain-compensation curve was kept constant on the ultrasound instrument and videodensity was measured from videotapes by an offline computer system. In the image of each cone, a 10 X 70 pixel operator-designated region of interest was chosen, and mean videodensity/pixel (256 level gray scale) was calculated. A logarithmic relation was found between scatterer concentration and videodensity. This excellent log relationship existed at all three gain levels studied: at 40% of maximum gain, r = .96; at 50% gain, r = .99; and at 60% gain, r = .99 (all p less than .001). Boluses of .5 ml sonicated Hypaque were then injected into the chamber and the contrast washout was recorded at ejection fractions of 20, 30, 40, 50, 60, and 70%. Bolus delivery and complete contrast mixing was attained except at ejection fraction 20%. Using the first two to four stroke videodensity decay the ejection fraction was calculated by estimating microbubble concentration decline using the logarithmic regression equation between videodensity and scatter concentration obtained during calibration.(ABSTRACT TRUNCATED AT 250 WORDS)