Pitfalls in creation of left atrial pressure-area relationships with automated border detection

Creation of pressure-area relationships (loops) with automated border detection (ABD) involves correction for the variable inherent delay in the ABD signal relative to the pressure recording. This article summarizes (1) the results of in vitro experiments performed to define the range of, and factors that might influence, the ABD delay; (2) the difficulties encountered in evaluating a thin-walled structure like the left atrium in the dog model; and (3) the solutions to some of the difficulties found. The in vitro experiments showed that the ABD delay relative to high-fidelity pressure recordings ranges from 20 to 34 msec and 35 to 57 msec at echocardiographic frame rates of 60/sec and 33/sec, respectively. The delay was not influenced significantly by the type of transducer used, distance from the target area, or size of the target area. The delay in the ABD signal, relative to the echocardiographic image, ranges from nil to less than one frame duration, whereas it is delayed one to two frame durations relative to the electrocardiogram processed by the imaging system. In the dog model, inclusion of even small areas outside the left atrium rendered curves with apparent physiologic contour but inappropriately long delays of 90 to 130 msec. To exclude areas outside the left atrial cavity, time-gain compensation and lateral gain compensation were used much more extensively than during left ventricular ABD recording. By changing the type of sonomicrometers used in our experiments, we were able to record simultaneously ABD and ultrasonic crystal data. However, both spontaneous contrast originating from a right-sided heart bypass pump and electronic noise from the electrocautery severely interferred with ABD recording.