压缩感知在医学超声成像中的仿真应用研究

Simulation of the application of compressive sensing to medical ultrasound imaging

为了解决医学超声成像系统中面临的采样率高,数据量大的问题,提出将压缩感知理论方法用于医学超声成像。首先建立了超声信号在时域的稀疏表达模型,然后利用模拟信息转换器对信号进行稀疏采样,最后使用最优化方法完成回波信号重建,利用合成发射孔径方式完成最终超声成像。为了验证算法的有效性,利用Field II对点目标以及复杂组织目标进行了仿真实验,在均方误差、分辨率、对比度以及成像质量上与常规成像结果对比分析。结果表明,采用1/2奈奎斯特采样频率,以30%原始数据所完成的成像仍然可保证良好的图像质量。采用压缩感知理论可以大幅度降低医学超声系统的采样率及总数据量。

In order to lower the sampling rate and to reduce the huge amount of data in imaging of synthetic transmitting aperture, a medical ultrasound imaging method based on compressive sensing is presented. Firstly, the sparsity of ul- trasonic echo signal in time-domain is verified. Then, the echo signal is sparely sampled by an Analog-to-Information converter. Finally, the echo signal is reconstructed by solving an optimization problem. Experiments for point target and complex tissue target are used to verify the proposed method. The RMS errors, resolutions, contrasts and image qualities of the reconstruct image and the original image are compared. The results show that ultrasound imaging can be im- plemented with a sampling rate below Nyquist frequency and a data amount of only 30% without reducing the quality of image.