On-line automated border detection for echocardiographic quantification of right ventricular size and function in children.

Abstract

Rapid, accurate assessment of right ventricular (RV) size is important for the management of children with congenital heart disease. The usefulness of the Acoustic Quantification system of automated border detection (ABD) and on-line quantification (AQ) for assessment of RV size was tested in 36 children. AQ data were compared to "corrected AQ" measurements (after correction for cavity areas erroneously included in the region of interest) required for AQ. Furthermore, the influence of necessary changes to gain settings was tested in "lateral gain control" (LGC) images obtained by removal of ABD overlays. All results were compared to conventional echocardiography (echo), and agreement with magnetic resonance imaging (MRI) RV areas was assessed. Systematic differences (+/-) limits of agreement with MRI (transverse plane) for conventional echo and AQ (apical four-chamber view) were as follows: end-diastolic -0.8 +/- 3.8 (conventional echo) versus -1.7 +/- 4.6 (AQ) cm2/m2 (p < 0.001); end-systolic -1.3 +/- 3.2 versus -4.9 +/- 5.8 (AQ) cm2/m2 (p < 0.001); fractional area change 7.8 +/- 17.0% versus 26.9 +/- 31.4% (AQ) (p < 0.001). Differences between conventional echo, LGC, and corrected AQ areas were not statistically significant. The best agreement between MRI and echocardiography was with conventional echo. We conclude that automated border detection of the RV can be performed successfully with the AQ system at a fixed point in the cardiac cycle. For adequate assessment of RV function manual corrections of online AQ results are still required, which results in an important reduction of the time gain of on-line quantification.