Automated on-line quantification of left ventricular ejection fraction by echo.

ABSTRACT

An automated border detection system based on the acoustic back scatter property of ultrasound has been recently developed. This provides beat-to-beat values for the left ventricular ejection fraction (EF). This study assesses the reliability of this method when compared with the angiocardiographic EF, and manual 2D-echo-derived EF based on the modified Simpson method. Angiocardiographically-derived EF was used as the gold standard. In a total of 49 patients the angiocardiographic EF was found to be 64.49 +/- 15%. It showed excellent correlation with the automated border detection-derived EF (r = 0.91). The automated border detection-derived EF also showed excellent correlation with the manual 2D-echo method (r = 0.9). In addition, automated border detection requires less time than the manually drawn method (40.9 +/- 7.7 sec versus 58.6 +/- 7.5 sec; p < 0.001). We conclude that the acoustic quantification of the EF correlates well with the angiocardiographically-derived EF and is less time-consuming.