【作者】 肖杨；

【导师】 余建国；

【作者基本信息】 复旦大学 ， 医学电子学， 2010， 博士

【摘要】 医学超声检测系统具有无损、价廉、实时的优点,被广泛应用于人体血流动力学参数测量领域,对生理研究和疾病诊断都具有重要价值。近年来,随着工程电子、信号处理、半导体集成电路等现代高新技术的迅速发展,提高血流测速分辨力、改善回波信噪比、增强系统的灵活性,实现系统的智能化和小型化成为医学超声检测领域的研究热点。论文深入研究了超声血流检测中涉及到的若干关键性问题,提出了一套完整的系统解决方案。采用软件无线电思想和编码激励技术来改进传统超声系统的硬件平台,并配以高性能的流速估计方法,用以实现血流速度剖面的定量测量。论文的创新性工作包括了硬件平台和软件算法两个部分的内容,主要概括如下：1)将软件无线电思想引入到系统平台构建中,将高速模数转换推进到超声接收电路的最前端,直接对未经解调的超声射频回波信号进行采样量化,使得大部分的系统功能完全由软件实现。同时,采用了现代集成电路技术和数字信号处理技术的最新成果,提出了一个稳定的、通用的全数字化硬件平台设计方案,实现了超声信号的发射、接收、采样量化和后续处理。2)将编码激励技术引入到超声发射信号的改进中,对适用于血流检测的编码特性进行了深入剖析,从编码优选的角度出发,提出一种ZCWC码序列组,该组序列具有区域最理想的非周期自相关性和互相关性,在零相关窗内可以完全消除旁瓣。论文给出了ZCWC码序列组的构造方法并分析了采用该码测量血流速度剖面的原理,通过单反射面回波实验、混合编码回波分离实验和模拟血流测量实验验证了该码的优良特性。3)依据超声血流散射回波信号的形成机理,提出一种更精确的分布式速度回波信号模型来描述散射子运动的时延效应,首次引入了分布参数的概念来反映测量单元内流速的分布状况。4)针对传统窄带流速估计方法会受到中心频率偏移和短脉冲频谱展宽导致精度下降的缺点,提出了一种基于频谱特征匹配的2-DFST算法。直接对携带着原始运动信息的超声射频回波序列进行二维傅立叶变换,利用发射信号生成动态特征模板,采用简化的相关系数作为相似性度量准则进行跟踪匹配,由匹配成功后的特征点坐标得到宽带回波信号的中心频率和多普勒频移的估计,计算出流速测量值。5)针对高轴向分辨率的实时系统需要利用小样本进行有效快速的估计的特点,提出了一种基于分布式速度回波信号模型的TVJE算法,将传统的单一速度估计问题扩展成中心速度和分布参数的二维联合估计。根据表征速度扩展程度的分布参数估计值动态调整采样容积的大小,利用扩充后的样本点估计出最终的流速剖面。实际应用中,血流散射回波信号的随机扰动对测量结果的影响超过算法本身性能的影响,尤其是在样本数较少的情况下。TVJE法首次提出分布参数作为估计的置信度准则,从另一个角度提供了提高流速估计质量的新思路。该方法可有效改善估计算法在空间非平稳信号环境下的表现,提高了利用小样本测量流速剖面的精确度和稳健度。完成了硬件平台的搭建和软件算法的调试后,采用该系统进行了模拟血流测量实验研究。实验结果表明系统能够实现仿血液的速度剖面的定量测量,实验结果与理论分析基本一致,尤其在小样本数据时能获得良好的测量效果。更多还原

【Abstract】 The medical ultrasonic-inspection system has been widely applied in the hemodynamic parameters measurement due to its merits of noninvasiveness, low cost and real-time processing, which has proved to be a valuable tool for pathophysiology research and clinical diagnosis of vascular disease. With the rapid development of modern techniques, several topics recently become the hotspots in the research of the system, which include improving the measuring resolution, raising the signal to noise ratio of echoes, enhancing the sensitivity of the system, as well as designing the system with intelligentization and miniaturization.In this dissertation, several crucial problems in the measurement of blood velocity are studied and analyzed, and a complete intellectual system solution is proposed. In order to improve the traditional ultrasound measurement system, the idea of software radio (SR) and coded excitation technique are introduced into the hardware platform design. In addition, some new methods with improved performance are developed to flow velocity estimation. The research including hardware and software achieves the following innovative points:1) The SR idea is introduced into building the system platform. The high speed A/D converter is used to sample the ultrasonic radio frequency (RF) echo signals directly and the most of functions are accomplished by software. With the integrated circuit technique and the digital signal processing, a stable and all-purpose hardware platform is carried out for ultrasonic signals transmitting, receiving, sampling and post-processing.2) Coded excitation is applied to improve the ultrasonic transmit signal. After discoursing learnedly on the characteristics of the known codes which have been adapt to blood velocity measurement, a zero correlation window complementary (ZCWC) coded sequence is firstly presented for its prefect local autocoiTelation and cross correlation. The structure of ZCWC code and its good feature in measuring blood velocity profile are studied and analyzed. It is proved that ZCWC code provides higher performance by the mirror reflection experiment and the mix codes separation experiment.3) According to the ultrasonic echo signal mechanism of the human blood, a distributed model which represents the time delay effect of the scatters movement more accurately than the conventional model is setup. The distributed parameter is firstly utilized in the model, which is able to denote velocity extension in sampling unit.4) To cope with the estimation error caused by central frequency downshift and the wide frequency band, a two-dimensional frequency shift tracing (2-D FST) algorithm based on the spectral feature matching is proposed. The first step is to represent the received RF echo sequence signals in the two-dimensional spectrum by applying the 2-D FFT and then analyze the resulting spectral feature. After using the transmit signals to build dynamic templates, a simplified correlation coefficient tracing strategy is presented to acquire the coordinate of feature points. Finally, the velocity could be estimated by calculating the central frequency shift and the Doppler frequency shift.5)To improve the quality of the estimation in small sampling data situation, a time delay vector joint estimation (TVJE) method based on the distributed model motioned above is used to turn the conventional velocity estimation into the joint estimation of the central velocity and a distributed parameter. Then the velocity estimation is obtained from the central velocity and its validity is represented by the distributed parameter which denotes the velocity extension. After adjusting the sampling volume by the distributed parameter dynamically, the final velocity profile can be acquired.After building the hardware platform and debugging the software algorithm, the system was tested in the velocity measurement of simulated blood flow. It is proven that the system is capable of measuring the blood flow velocity profile, and the measurements showed a good agreement with the theoretical analysis. The system could provide good blood flow velocity measurements, especially under the circumstance of small sampling data.更多还原