基于自适应局部余弦变换的正交多普勒超声信号降噪方法的研究

Quadrature Doppler Ultrasound Signal Denoising Based on Adapted Local Cosine Transform

多普勒超声信号的谱图已经被广泛用于医疗诊断。来自系统内部的噪声及外部的干扰会产生附加的频谱成分,从而影响谱图的主观分析及进一步的定量分析。为抑制噪声的影响,本文提出利用一种新的基于自适应局部余弦变换和非负G arrote取阈值的方法对正交多普勒超声信号进行降噪。首先,由正交信号提取正向和逆向血流信息;然后对其分别进行降噪;最后利用H ilbert变换进行重构得到真实信号的估计。在仿真研究中,采用平均频率波形和谱宽波形的估计精度作为性能改善的指标。结果表明这种方法优于基于小波变换的降噪方法,特别是在低信噪比情况下。

The spectrogram of Doppler ultrasound signal has been widely used in clinical diagnosis. The additional frequency components arising from internal or external noise to the system will produce adverse effects on its subjective and quantitative analysis. A novel approach based on the adapted local cosine transform and the non-negative Garrote thresholding method was proposed to remove noise from quadrature Doppler signal. At first, the directional information was extracted from the quadrature signal. And then the denoising method based on the adapted local cosine transform is performed on the forward and backward flow signals, respectively. At last, the estimated signal was reconstructed from the denoised signals using Hilbert transform. In the simulation study, both the mean frequency and spectral width waveform were studied for the denoised signal. The simulation results had shown that this approach was superior to that based on the wavelet transform, especially under low SNR conditions.