Quality versus framerate in color flow imaging: an experimental study based on offline processing of RF-signals recorded from patients

Digital RF data from a color flow imaging scanner has been recorded in real time for offline processing. In this study color flow imaging data with low framerate (16 frames/sec) was recorded from a 3.5 cm wide region in the left ventricle. The effect of increasing the framerate was studied by reducing the number of pulses per beam (packet size) from 30 down to 4 using the same data. Clutter echoes were removed by a polynomial regression filter with variable order, and the velocity was estimated by the standard autocorrelation algorithm. Acceptable image quality was obtained with packetsize down to 4, corresponding to a framerate of 102 frames/sec. Velocity values were compared in a region with fairly uniform velocity field close to the mitral valve in diastole. Velocities obtained with packet size 30 was used as a reference, and bias and standard deviation were calculated for lower packet sizes. For packet size 8 and larger, the bias and standard deviation were below 0.12 cm/sec, and 1.54 cm/sec, respectively. By spatial averaging, similarity results were obtained for packet size 4. It is concluded that the accuracy of the velocity estimates was acceptable at framerates up to 102 frames/sec.