改进的最小方差算法在超声成像中的应用

An improved minimum variance algorithm in ultrasound imaging system

为了提高最小方差超声成像算法的分辨率、对比度以及对噪声的鲁棒性,提出一种改进的最小方差成像算法。该方法首先基于回波信号中期望信号与噪声信号的可分离性将信号划分为期望信号和噪声信号,然后根据最小方差原理,求出加权向量使期望信号功率最小,同时,为了增加算法对噪声的鲁棒性,对信号方向向量增加一对约束条件,进一步提高图像质量。在全发全收和合成孔径模式下对点目标和吸声斑进行仿真,结果表明所提算法在全发全收模式下,-6 dB处分辨率在最小方差基础上提高了1倍左右,在合成孔径模式下,对比度在特征空间最小方差算法基础上提高了8 dB,且远优于传统延时叠加算法。最后通过实验进一步表明改进的最小方差算法图像在分辨率、对比度及对噪声的鲁棒性等方面表现更优,可以有效的改善超声图像的质量。

In order to improve the resolution, contrast ratio and the robustness against the noise of the minimum variance （MV） method, an improved minimum variance （IMV） method was introduced. The received signal was divided into desired signal and noise signal at first, then the adaptive weighting vectors were obtained by minimizing the output power of the desired signal. In addition, a pair of constrained conditions were added to the steering vector to further enhance the robustness ability against the noise. Points target and cyst phm, tom were simulated under both the plane wave mode and the synthetic mode. The simulation results showed that the lateral resolution of the proposed method was improved by a factor of 2 in plane wave mode, and the contrast ratio was increased over 8 dB in synthetic aperture mode on the basis of eigenspace-based minimum variance （ESBMV）. Furthermore, the proposed method was far better than traditional delay and sum （DAS） algorithm. Finally, the experiment was conducted based on the real ultrasound data. The results indicated that the proposed method had a better resolution, contrast ratio and was more robust to the noise, which showed its potential application to enhance ultrasound imaging quality.