基于分数阶的医学超声图像去噪技术研究

发布时间：2018-06-07 20:54

本文选题：医学超声图像 + 分数阶微分　；参考：《昆明理工大学》2017年硕士论文

【摘要】：因为医学超声成像具有低成本、高效率、实时、方便安全的特点,所以它已逐渐成为最重要的诊断工具之一。然而粒状结构叠加的斑点噪声广泛存在于医学超声B模式图像中,其噪声表现为像素灰度值的突变,模糊了或掩盖了图像的细节,但是很多临床诊断至关重要的是医学超声图像的细节信息。因而,斑点是医学超声B模式图像低对比度的主要原因之一,它可以被看作是一个噪声源,应当抑制。在临床医学诊断中超声成像技术应用很广泛,但是因为受系统成像机制的影响,超声成像技术形成的图像比较容易形成斑点噪声,这将导致后期图像分析工作的要求难以满足。为了提高图像对比度,医学超声图像斑点去噪技术已经得到发展,可以保留图像的边缘,抑制斑点噪声,使得临床医生能准确地识别和分析病变区域。因此,医学超声图像去噪技术研究对临床诊断具有重要的现实意义。本文首先简单介绍了超声成像原理、超声B模式成像、超声斑点噪声模型以及去斑性能评价标准等,然后分析了几种经典的超声斑点抑制算法,如斑点抑制各项异性扩散方法(SRAD)、细节保留各项异性扩散方法(DPAD)、双边滤波(BF)和P-M模型。针对医学超声图像中降低图像质量及导致诊断困难的斑点噪声的颗粒模型特征,本文提出了基于分数阶的医学超声图像降噪方法,为了保留更多的纹理信息,基于具有k(阻止扩散的梯度阈值)和v(分数阶阶数)平衡关系的分数阶偏微分方程的图像去噪模型被构建,它有效结合了分数微积分理论和偏微分方程方法,并且通过分数阶掩模算子实现了它的数值计算方法。与其他三种传统的超声去噪方法(P-M模型、SRAD和DPAD)相比,本文提出的基于分数阶的各向异性扩散算法(FAD),在去除斑点噪声的同时保留组织结构方面更有优势。在产生相同实验结果的条件下,本文提出的算法比双边滤波(BF)运行速度更快。医学超声体模图像和人体成像实验表明,基于分数阶的各向异性扩散方法可以提高组织的信噪比和超声B模式图像的质量。
[Abstract]:Medical ultrasound imaging has become one of the most important diagnostic tools because of its advantages of low cost, high efficiency, real-time, convenient and safe. However, speckle noise superimposed by granular structures is widely found in medical ultrasound B-mode images. The noise appears as a mutation of pixel gray value, which obscures or conceals the details of the image. But many clinical diagnoses are critical to the details of medical ultrasound images. Therefore, speckle is one of the main reasons for the low contrast of B-mode medical ultrasound images. It can be regarded as a noise source and should be suppressed. Ultrasonic imaging technology is widely used in clinical medical diagnosis, but because of the influence of system imaging mechanism, the image formed by ultrasonic imaging technology is easy to form speckle noise, which will lead to the later stage of image analysis difficult to meet the requirements. In order to improve image contrast, medical ultrasound image speckle denoising technology has been developed, which can preserve the edge of the image, suppress speckle noise, and enable clinicians to accurately identify and analyze the lesion area. Therefore, the study of medical ultrasound image denoising technology has important practical significance for clinical diagnosis. In this paper, the principle of ultrasonic imaging, ultrasonic B-mode imaging, ultrasonic speckle noise model and evaluation criteria of speckle removal are briefly introduced, and then several classical algorithms for ultrasonic speckle suppression are analyzed. For example, speckle suppression of heterosexual diffusion methods (SRADX), detail retention of various heterosexual diffusion methods (DPADX), bilateral filtering (BF) and P-M model. In order to preserve more texture information, a fractional-order image de-noising method is proposed to reduce the image quality and the speckle noise in medical ultrasound images. The image denoising model of fractional partial differential equation with k (gradient threshold to prevent diffusion) and v (fractional order) equilibrium is constructed, which effectively combines fractional calculus theory with partial differential equation method. The numerical calculation method is realized by fractional mask operator. Compared with the other three traditional ultrasonic denoising methods, SRAD and DPAD, the fractional order anisotropic diffusion algorithm proposed in this paper has more advantages in removing speckle noise while preserving the structure of the tissue. Under the same experimental results, the proposed algorithm is faster than the two-sided filter. Medical ultrasound phantom images and human body imaging experiments show that fractional order anisotropic diffusion method can improve the signal-to-noise ratio of tissues and the quality of B-mode ultrasound images.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R310;TP391.41

【参考文献】