Optimization of Doppler echocardiographic velocity measurements using an automatic contour detection method

Intra- and interobserver variability in Doppler echocardiographic velocity measurements (DEVM) is a significant issue. Indeed, imprecisions of DEVM can lead to diagnostic errors, particularly in the quantification of the severity of heart valve dysfunctions. To reduce the variability and rapidity of DEVM, we have developed an automatic method of Doppler velocity wave contour detection, based on active contour models. To validate our new method, results obtained with this method were compared with those obtained manually by two experienced echocardiographers on Doppler echocardiographic images of left ventricular outflow tract and transvalvular flow velocity signals recorded in 30 patients with aortic or mitral stenosis, 20 with normal sinus rhythm and 10 with atrial fibrillation. We focused on the three essential variables that are measured routinely using Doppler echocardiography in the clinical setting: the maximum velocity (Vmax), the mean velocity (Vmean) and the velocity-time integral (VTI). Comparison between the two methods has shown a very good agreement. A small bias value was found between the two methods (between -3.9% and 0.5% for Vmax, between -4.6% and -1.4% for Vmean and between -3.6% and 4.4% for VTI). Moreover, the computation time was short, approximately 5 s. This new method applied to DEVM could, therefore, provide a useful tool to eliminate the intra- and interobserver variabilities associated with DEVM and thereby to improve the accuracy of the diagnosis of cardiovascular disease. This automatic method could also allow the echocardiographer to realize these measurements within a much shorter period of time compared with the standard manual tracing method. From a practical point of view, the model developed can be easily implemented in a standard echocardiographic system.