Rajaonah J C, Dousse B, Meister J J
Lab. de Genie Med., Ecole Polytech. Federale de Lausanne.
IEEE Trans Ultrason Ferroelectr Freq Control. 1994;41(6):812-9. doi: 10.1109/58.330262.
In the presence of noise, any wall filter induces a bias on the mean Doppler frequency estimated by the algorithm based on the calculation of the phase of the autocorrelation function at lag 1. In this paper, it is shown that the bias results from the nonzero value of the autocorrelation function of the filtered noise at lag 1. A general method for compensating the bias was then deduced. It consists of a preliminary estimation of the filtered noise contribution and its subtraction from the autocorrelation function of the Doppler signal. The method is independent of the nature of the noise and wall filter. An evaluation of this method on simulated Doppler signals and on measurements from a moving-string phantom showed that effective compensation of the bias was obtained, but at the expense of a higher variance. Taking into account both the bias and the variance, however, it was shown that this method offers an improvement over the noncompensated method. Finally, the performance of this method was demonstrated using in vivo measurements taken from a human aorta.
在存在噪声的情况下,任何壁滤波器都会对基于滞后1处自相关函数相位计算的算法所估计的平均多普勒频率产生偏差。本文表明,该偏差源于滞后1处滤波后噪声的自相关函数的非零值。然后推导了一种补偿该偏差的通用方法。它包括对滤波后噪声贡献的初步估计,并将其从多普勒信号的自相关函数中减去。该方法与噪声和壁滤波器的性质无关。对模拟多普勒信号和移动弦体模测量结果进行的该方法评估表明,虽然获得了偏差的有效补偿,但代价是方差更高。然而,综合考虑偏差和方差后表明,该方法比未补偿方法有所改进。最后,使用从人体主动脉获取的体内测量结果证明了该方法的性能。