Vessel wall tracking based on the modified autocorrelation estimator

Abstract

In 16 subjects (age 26-74) we recorded radio-frequency (RF) data of the Carotid artery with a Vivid7 scanner (GE Vingmed Ultrasound, Horten, Norway) and Carotid diameters with WallTrack (Pie Medical, Maastricht, The Netherlands). The RF data were stored as complex demodulated radio-frequency data (IQ data) and transferred to a PC for estimation of diameter curves. For the prototype system, diameter curves are derived using a modified autocorrelation method, in contrast to Walltrack where the complex cross-correlation method is applied. For all datasets, 2-5 beats were averaged to calculate representative one-cycle curves. The diameter distension /spl Delta/D as obtained with both systems was in the range 0.27-0.71 mm, and was highly correlated (r = 0.91, p < 0.002, y = 0.99x + 0.03) with a mean difference of -0.03 mm and limits of agreement (mean difference /spl plusmn/ 2*SD of the differences) of -0.15-0.09 mm. The relative distension /spl Delta/D/D ranged from 3.5 to 11.1% and had a correlation coefficient of r = 0.92 (p < 0.002, y = 0.94x + 0.75). The mean difference of /spl Delta/D/D was -0.4%, and the limits of agreement were -2.2-1.4%. In conclusion, this study indicates that it is adequate to use the modified autocorrelation method for vessel wall tracking.