【作者】 刘琳波；

【导师】 郁道银；

【作者基本信息】 天津大学 ， 光学工程， 2004， 硕士

【摘要】 在对消化道病变的诊断当中，单纯根据粘膜表面的形态变化来推断病变情况的方法，存在一定的主观性和局限性。内窥镜超声成像系统可以通过超声扫描，获得较为清晰、准确的消化道器官的断层图像，为医生的对症施治提供了客观的依据。相对于其他成像方式，超声成像具有无创伤、无电离辐射、重复性好等优点。内窥镜超声成像系统的研制工作主要包括探头设计、图像采集和图像处理三个部分的内容。受内窥镜活检通道尺寸的限制，探头的设计面临诸多难题，其中包括微型驱动源和微型高频超声换能器设计。在B型超声机械扇形扫描成像当中，原始图像信号是以极坐标的形式输出的，对图像进行实时坐标变换和漏点插补是图像预处理所要解决的主要问题。通过采用先进的微型超声马达和马达前置的设计方案，本文设计了无电磁干扰、长寿命、匀速、大转矩的驱动源。在换能器设计当中采用复合材料和球面单源的结构形式，为宽带、高灵敏度内窥镜超声探头的研制奠定了良好的基础。本文采用数字坐标旋转(CORDIC)算法将极坐标图像转化为直角坐标图像，并采用三次样条插值的方法对漏点进行弥补。在算法实现过程当中，充分利用FPGA的空间计算能力，设计了流水线式并行处理结构，获得了很高的处理速度。本文搭建了以复杂可编程逻辑器件(CPLD)为核心的硬件系统建构，并在该系统当中利用Verilog HDL实现了上述图像处理任务。仿真实验证实，该方法能在保证图像质量的基础上满足实时性的要求。本系统采用512×512大小、帧频15Hz的超声图像，图像从输入到输出的时间延时是一帧时间，相当于67ms，因此，观察超声图像时，人眼基本感觉不到延时的存在，达到了实时性的设计要求。系统的误差来源和图像处理精度的分析结果证明，图像处理精度较高，失真较小。最后，总结了系统的特点，提出了改进的措施和方法。更多还原

【Abstract】 To diagnose disease of the alimentary canal, changes of mucous membrane suffice fail to provide sufficient information to make out an objective diagnosis of the pathological changes. However, Endoscopic Ultrasonography System (EUS) can help obtain clear and accurate ultrasonic tomography images, and thus provide objective and sufficient information for diagnosis and therapy; In addition, compared to other imaging methods, the ultrasonic imaging are non-invasive, free of ionization and radiation, good at repeatability. The procedures to develop the EUS include probe design, image acquisition and image pre-processing. Due to limited size of biopsy channel, the probe design is confronted with many difficulties, esp. micro-motor and high frequency micro-transducer; In the B-mode sector scan ultrasonic imaging system, the raw image is in polar coordinates, therefore, it is main tasks of image pre-processing to achieve image coordinate transform and interpolation. Through introducing the advanced micro-ultrasonic motor and the front motor design into mini-probe design, an EMI free, steady, long life motor with large torque is obtained; the Piezo-composite and single element spherical transducer are employed to achieve the high bandwidth, high sensitivity of the EUS probe.The CORDIC algorithm is adapted to convert the polar coordinate image to the rectangular coordinate image, while cubic spline interpolation is used to interpolate the vacant pixels. To achieve the real-time feature, the spatial computation ability of FPGA is fully explored to enhance the parallel, pipelined computing. Based on FPGA and Verilog HDL, the proposed algorithms are complied and verified. Simulation results performed in Matlab and Quartus II environments prove that the proposed algorithms can achieve satisfying image quality and real-time feature.This system uses ultrasonic video with image size of 512×512 and frame frequency of 15Hz. The time delay between image input and output is a frame cycle or around 67ms, so that human eyes hardly can perceive the delay and consequently it is a real-time process. It is proved by error source analysis and image processing accuracy analysis that the proposed algorithms have a satisfying accuracy. In the end, features of the whole system are summarized and further work for improvement is indicated.更多还原