基于FPGA的超声成像核心算法研究

发布时间：2018-06-27 15:57

本文选题：FPGA + 全数字超声诊断仪　；参考：《华中科技大学》2012年硕士论文

【摘要】：近年来计算机数字技术和集成电路的不断发展，促使全数字化超声诊断仪的进一步改进和完善。在硬件FPGA环境下对全数字超声诊断仪的内部核心算法进一步改进，旨在从本质上为医学超声诊断提供更好质量的超声成像，同时也为研究三维超声医学奠定基础。本文在简单介绍了全数字超声诊断仪的基本框架之后，首先基于超声信号衰减随扫描深度增加而增加这一原理，设计了符合要求的随着扫描深度增加中心频率下移的数字动态滤波器；针对FPGA环境下实现平方和开方运算带来的误差和运算复杂度的问题，将数字正交检波后的幅度提取和对数压缩相结合，省去了开方运算，减少了对数压缩消耗的资源。对每个单通道数据进行下采样，从而得到每个单通道数字RF回波信号的8位数据。接着重点介绍了数字扫描转换(DSC)模块里面的CORDIC坐标转换和二维线性插值算法，对DSC中CORDIC坐标转换算法的精确度进行改进，设计了显示平面、图像有效显示区域的坐标产生模块和节省资源的图像象素存储方式,实现了从一维RF回波信号到二维数字扇型扫描图像的成像过程,同时，对本设计FPGA得到的结果和MATLAB下独立产生的参考图像通过直观的残差图和客观评价参数SSIM(结构相似度)、PSNR(峰值信噪比)进行对比评价，验证了本设计的正确性和可行性。最后，对整个工作进行了总结，以及对本设计中存在的一些不足提出后期工作展望。
[Abstract]:In recent years, the continuous development of computer digital technology and integrated circuit, promote the further improvement and improvement of all digital ultrasonic diagnostic instrument. In the hardware FPGA environment, the internal core algorithm of all-digital ultrasonic diagnostic instrument is further improved, in order to provide better ultrasound imaging for medical ultrasonic diagnosis in essence, and also lay a foundation for the study of three-dimensional ultrasound medicine. After a brief introduction of the basic framework of the full digital ultrasonic diagnostic instrument, this paper firstly bases on the principle that the attenuation of ultrasonic signal increases with the depth of scanning. A digital dynamic filter with decreasing center frequency with increasing scanning depth is designed, and the error and computational complexity caused by square sum square operation in FPGA environment are analyzed. The amplitude extraction after digital orthogonal detection is combined with logarithmic compression, which saves the square operation and reduces the consumption of logarithmic compression. The 8 bit data of each single channel digital RF echo signal is obtained by downsampling each single channel data. Secondly, the Cordic coordinate conversion and two-dimensional linear interpolation algorithm in DSC module are introduced. The accuracy of Cordic coordinate conversion algorithm in DSC is improved, and the display plane is designed. The coordinate generation module of the effective display area and the saving of image pixel storage method are used to realize the imaging process from one dimension RF echo signal to two dimensional digital fan scan image, at the same time, The results obtained from FPGA and the reference images generated independently under MATLAB are compared and evaluated by visual residuals and objective evaluation parameters (SSIM) and PSNR (peak signal-to-noise ratio), which verify the correctness and feasibility of the design. Finally, the whole work is summarized, and some shortcomings in this design are discussed.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R445.1

【参考文献】