超声弹性成像技术的研究与实现

发布时间：2018-04-25 23:37

本文选题：超声弹性成像 + 光流跟踪　；参考：《北京协和医学院》2017年硕士论文

【摘要】：弹性成像技术可以提供人体组织的弹性力学相关数据描述组织的软硬程度,在疾病诊断和检测中起到了不可忽视的重要作用。在这项技术应用于临床以前,医生一般采用触诊方式检测组织硬度的变化,具有主观、重复性差、诊断深度浅等缺点,不利于某些疾病的诊断和治疗。弹性成像的出现填补了传统的成像技术不能区分结构相似但软硬不同的组织的缺点,与超声成像结合可以同时提供形态学和弹性信息,为疾病检查和诊断提供了新的方法,补充了力学特性和成像参量。目前,超声弹性成像技术已成为医学成像范围内的热门研究之一。本文的研究对象是静态/准静态超声弹性成像,第一部分是一维位移和应变估计,超声弹性成像中的位移估计大致分为基于时域和基于频域的位移估计两大类,其核心为求解组织受到外部激励前后相应射频信号窗的时移,通过时域延伸以及幅频特性校正等手段对超声射频信号预处理,提高信号间的相关性。最后对求得的一维位移进行卡尔曼滤波,并用卷积算子计算求解应变估计。本文的第二部分是在传统的光流法的基础上,提出了一种基于矢量场预估的二维应变估计方法,将光流特征点定为基于包络线峰值的特征指纹信息,这样处理后使用光流跟踪时可以减少误差,实现超声射频信号的二维位移估计,通过设计特定的卷积算子对二维位移场快速差分求解得到相应的应变估计结果。本文的第三部分是实验的设计与分析,详细介绍了用有限元分析软件COMSOL和声场仿真软件FIELD Ⅱ生成仿真超声射频信号的方法,并用不同的四种模型对提出的方法进行验证,与一维位移估计相比,在二维相关性计算之前对组织整体位移进行预估处理,能有效提升计算速度与准确性。
[Abstract]:Elastography can provide elastic mechanics data of human tissue to describe the soft and hard degree of tissue, which plays an important role in the diagnosis and detection of disease. Before this technique is applied to the clinic, doctors generally use palpation to detect the changes of tissue hardness, which is subjective, poor repeatability and shallow diagnosis. The emergence of elastography fills the shortcomings of traditional imaging techniques that can not distinguish structures with similar structures but with different soft and hard structures. The combination of ultrasound imaging can provide morphological and elastic information at the same time, and provides a new method for disease detection and diagnosis, supplemented by mechanical properties and imaging. At present, ultrasonic elastography has become one of the most popular studies in the range of medical imaging. The object of this paper is static / quasi-static ultrasound elastography. The first part is one dimension displacement and strain estimation. The displacement estimation in Ultrasonic Elastic imaging is roughly divided into two categories based on time domain and frequency domain based displacement estimation. In order to solve the time shift of the corresponding radio frequency signal window before and after the external excitation, the heart is preprocessed by the time domain extension and amplitude frequency characteristic correction to improve the correlation between the signals. Finally, the Calman filter is used to obtain the one-dimensional displacement, and the second part of this paper is calculated by the convolution operator. On the basis of the traditional optical flow method, a two-dimensional strain estimation method based on the vector field prediction is proposed. The feature point of the optical flow is defined as the characteristic fingerprint information based on the peak of the envelope line. In this way, the error can be reduced when the light current is traced, and the two-dimensional displacement estimation of the ultrasonic radio frequency signal is realized, and the specific volume calculation is designed by designing a specific volume. The third part of this paper is the design and analysis of the experiment. The method of generating the emulation of the ultrasonic radio frequency signal by the finite element analysis software COMSOL and the sound field simulation software FIELD II is introduced in detail, and the proposed method is verified with four different models. Compared with the one-dimensional displacement estimation, the prediction of the overall displacement of the tissue before the calculation of the two-dimensional correlation can effectively improve the speed and accuracy of the calculation.

【学位授予单位】：北京协和医学院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41;R445.1

【相似文献】