Ultrasonic Speckle Tracking with an Adaptive Frame Interval for the Measurement of Blood Flow Velocities

Abstract

Blood flow velocity profile (BFVP) is commonly used to calculate hemodynamic parameters. Traditional speckle tracking (TST) estimates the displacements of image blocks between adjacent frame images to measure the BFVP. However, quantized errors caused by the transformation from image pixels to physical distances negatively impact on the accuracy of the BFVP measurement. In the present study, an improved speckle tracking (IST) is proposed to adaptively determine the frame intervals between two frame images based on the correlation of speckles. Firstly, the maximum normalized correlation coefficients (MNCCs) are calculated by finding the matches of the kernel blocks in subsequent comparison frame images (CFIs) after the reference frame image (RFI). Then, the largest frame interval with the MNCC greater than 0.6, is chosen to determine the velocity of the kernel block. In the experiments, the ultrasound transducer is used to scan a vessel-mimicking phantom with the peak velocities of 0.1 m/s and 0.2 m/s, respectively. The echo signals are off-line processed to yield B-mode images. The normalized root mean square errors of the TST-based results are 23.06±3.36 % and 19.92±3.08 %, and reduced to 8.36±1.86 % and 7.76±1.90 % by the IST method for the two different peak velocities. In conclusion, the IST method can improve the measurement accuracy of the BFVP, from which accurate diagnosis information could be hopefully acquired in clinics.