Intracardiac ultrasound determination of left ventricular volumes: in vitro and in vivo validation

Division of Cardiology, Hartford Hospital, University of Connecticut 06115.

OBJECTIVES. This study was designed to assess the feasibility of calculating left ventricular volumes using intracardiac ultrasound. BACKGROUND. Previous studies have validated transthoracic echocardiographic determinations of left ventricular volumes and have indicated the superiority of Simpson rule reconstruction algorithms. The feasibility of imaging the left ventricle with intracardiac ultrasound has also been demonstrated. METHODS. The determination of left ventricular volumes with Simpson rule reconstruction of intracardiac ultrasound images was evaluated in two phases. In vitro validation was performed in 29 animal hearts preserved in either a nondistended or distended state. Latex cast volumes were the reference standard. In vivo studies used 14 pigs, and compared intracardiac ultrasound volumes and ejection fraction with single-plane contrast angiographic values. A 12.5-MHz device was used to record short-axis images at 0.5-cm intervals. These were used to reconstruct the ventricle as a stack of cylindric elements using all imaged levels as well as sections recorded every 1 and 2 cm and at a single midventricular level. RESULTS. In the in vitro hearts, when all recorded sections were used, there was excellent agreement between intracardiac ultrasound and latex cast volumes (intracardiac ultrasound volume = 0.89 latex cast volume + 2.22, r = 0.95; intracardiac ultrasound volume = 0.97 latex cast volume + 0.91, r = 0.99) for nondistended and distended hearts, respectively. In vivo, there was again close correspondence between ultrasound and angiographic volumes (intracardiac ultrasound volume = 1.04 angiographic volume - 3.6, r = 0.91). The relation between intracardiac ultrasound and angiographic ejection fraction was fair (intracardiac ultrasound ejection fraction = 1.00 angiographic ejection fraction + 6.85, r = 0.69). Excellent correlations for the volumes were maintained as the number of cross sections was reduced to those recorded every 1 and 2 cm (r = 0.87 to 0.99). With a single midventricular site more variable but generally good correlations were obtained (r = 0.77 to 0.99). CONCLUSIONS. The application of Simpson rule reconstruction to short-axis images of the left ventricle obtained with intracardiac ultrasound provides accurate determination of left ventricular volumes in animal hearts. This technique may prove useful in the analysis of left ventricular structure and function.