医学内窥超声成像设备在获取图像时受到换能器尺寸等多种因素的影响, 使得超声图像对比度较低和噪声较大, 无法为医疗诊断提供清晰的影像依据。为此科学家们提出了多种处理方法, 但这些方法多为成像后处理算法, 实时性较差, 无法满足内窥超声系统实时成像的要求 (25 f/s)。针对以上问题, 本文设计了基于提升式小波变换的嵌入式超声内镜实时图像处理系统, 利用超声内窥系统环扫成像特点以及 FPGA流水线概念, 对每条扫描线的回波信号进行小波去噪, 再经过 CORDIC算法、插值处理后得到二维超声图像。本文利用自行搭建超声内镜实验系统对鸡肉组织进行环扫成像, 实验表明该系统成像速度可达 25 f/s, 信噪比提高了 3.8 dB, 从而验证了系统的可行性。

During acquiring images, the equipment is affected by the size of the transducer and other factors which make the images have low contrast and big noise, cannot provide clear imaging basis for medical diagnostic. Scientists have proposed a variety of methods. However, these methods are mostly after-imaging processing algorithms and having a poor real-time performance, unable to meet the real-time endoscopic ultrasound imaging requirements (25 frames/s). Therefore, this paper designs the embedded endoscopic ultrasound real-time image processing system based on lifting Wavelet Transform. And the system processes echo signals of each scan line with wavelet denoising, then obtains two-dimensional ultrasound images after CORDIC algorithm and interpolation processing using the imaging characteristics of circular scan and FPGA pipeline concept. Through building my own endoscopic ultrasound experimental system which makes the chicken tissue imaging use circular scan, the experiments show that the system imaging speed is up to 25 frames /s and SNR improves 3.8 dB, verifying the feasibility of the system.