基于期望最大化框架的医学超声图像去斑

Despeckling medical ultrasound images based on an expectation maximization framework

针对医学超声图像斑点噪声,提出一种基于期望最大化(EM)框架的去斑算法。先从超声I/Q图像中提取实部;然后从该实部图像中"盲估计"出系统的点扩散函数;最后利用EM算法,在维纳滤波和各向异性扩散间进行迭代,从而获得去斑后的超声图像。对不同信噪比的仿真图像和实际图像采用本文方法和现有方法进行比较实验,结果表明,采用本文方法可将超声图像的斑点信噪比和边界保留指数平均提高1.94和7.52倍,归一化均方差平均降低3.95倍,性能指标优于现有方法。

In view of inherent speckle noise in medical images,a de-speckling method was proposed based on an expectation maximization（EM） framework.Firstly,the real part was extracted from the Inphase/Quadrature（I/Q） ultrasound image.Then,the point spread function was blindly estimated from the real image.Lastly,based on the EM framework,an iterative algorithm alternating between Wiener filtering and anisotropic diffusion was exploited to produce de-speckled images.The comparison experiment was carried out on both simulated and in vitro ultrasound images using the proposed method and exited ones,respectively.It was shown that the proposed method averagely improved the speckle-signal-to-noise Ratio（S-SNR） and the edge preservation index（β） of I/Q images by the factor of 1.94 and 7.52.Meanwhile,it averagely reduced the normalized mean-squared error（NMSE） by the factor of 3.95.The simulation and in vitro results indicated that the proposed method had a better overall performance than exited ones.