【作者】 徐海东；

【导师】 周兆英；

【作者基本信息】 清华大学 ， 仪器科学与技术， 2004， 博士

【摘要】 超声无损测温是一个集超声学、生物医学、热学、电子学、计算机科学、图像学、数学等众多领域在内的多学科交叉的前沿课题，不仅有着巨大的应用需求，而且与 EIT、X-CT、MRI、微波等其他测温技术相比，拥有相当多的技术积累和优势。 本论文以全身灌注热疗治癌为背景，以后向散射超声无损测温为目标，从测温模型和算法、测温仪设计与实现、离体和动物实验等各方面，探索出有效的超声无损测温方案，并指出在拥有高性能超声测温仪、基于先验知识的高效时移算法的前提下，超声测温将在组织定征参数 Km较小的组织首先临床实用。 本论文全面分析了建立在生物组织离散随机介质模型下的后向散射超声时移、频移测温模型，确立了时移模型的优势地位。从图像处理分析的角度提出了一种新的二值图像模糊跟踪算法（BIFT），进行了理论推导，并与互相关算法（CCA）和自回归功率谱密度（AR-PSD）算法做了比较。与传统的 CCA 技术相比，BIFT 在空间分辨率和记忆力、计算量、算法收敛性和抗噪性上都有明显改善。BIFT 已成功地应用于超声时移测温仿真和离体实验。 为了得到丰富的超声原始射频回波信息，传统的超声诊断仪不能满足超声测温模型和算法对超声测温仪提出的高采样率、高信噪比的要求。因此本论文研制了基于 PC 的全数字化 A/B 型超声无损测温仪，包括超声探头的选择、测温仪整体考虑、全数字化超声、高采样率高增益高信噪比放大电路、频率自适应等关键技术，并给出了基于 A/B 两种不同探头的设计实例。本论文研制的超声测温仪可以满足时移、频移和能量模型等任何超声脉冲回波式测温系统的需要。 本论文通过数据仿真、水温实验、离体实验、组织定征以及动物实验等各个环节，提出了一套实用的超声无损测温临床方案。研究发现，与加热模式和目标尺寸等先验知识相关的时移拟合阶次、超声测温仪的采样率和信噪比、组织定征参数 Km 是直接影响温度评估精度的三大因素；提出了超声测温标记法，并考虑了临床运动补偿、人体热模型和人体解剖模型；给出了超声测温仪的最小采样帧率判据以及猪肝的 Km 值。更多还原

【Abstract】 Noninvasive temperature estimation using ultrasound is an advancedinterdisciplinary topic, which involves ultrasonics, biomedicine, thermology,electronics, computer science, image processing and analysis, mathematics, etc. It hasnot only great application requirement but also much more technology advantagesthan EIT, X-CT, MRI, Microwave, and other technologies of noninvasive temperatureestimation. This thesis is based on WBPH (Whole Body Perfusion Hyperthermia), andaimed at noninvasive temperature estimation using backscattered ultrasound. Itdeveloped a suit of effective scheme of noninvasive temperature estimation usingultrasound, through the aspects of ultrasound temperature model and algorithm,design and realization of ultrasound diagnostic apparatus, in vitro and in vivoexperiments, and etc. It points out that those tissue whose Km is smaller will beclinical used firstly in the noninvasive temperature estimation, if we have highperformance ultrasonic temperature measurement apparatus and excellent time-shiftalgorithm based on prior knowledge. This thesis fully analyzed ultrasound backscattered time-shift andfrequency-shift models for noninvasive temperature estimation, which are based onthe discrete random medium model of biotissue. We select the better model oftime-shift, and invent a new BIFT (Binary Image Fuzzy Tracking) algorithm from theview of image processing and analysis, present its theoretical basis, and compare itwith CCA (Cross Correlation Algorithm) and AR-PSD (Autoregressive PowerSpectrum Density). Comparing to conventional CCA, BIFT has better performancesat several aspects of spatial resolution and memory capacity, computationalcomplexity, algorithm convergence and anti-noise ability. BIFT has been usedsuccessfully in the simulation and in vitro tissue experiments of the time-shift modelfor noninvasive temperature estimation. For getting original ultrasonic RF echoes, the conventional ultrasonic diagnostic - III -<WP=6>英 文 摘 要apparatus can’t meet the requirements of high sampling frequency and highsignal-noise ratio for the ultrasound temperature estimation models and algorithm. Sothis thesis researched and developed the full-digital A/B mode ultrasonic temperaturemeasurement apparatus based on PC, including several key technologies such as theselection of ultrasound probe, overall consideration of diasonograph, full-digitalultrasound, the amplifying circuit with high sampling-frequency, high gain and highsignal-noise ratio, frequency self-adaptation circuit, and etc. At last this thesis givestwo designing examples based on the A/B mode probe respectively, which meet therequirements of any temperature measurement systems using ultrasonic pulse echoes,such as time-shift, frequency-shift, and power models. This thesis developed a suit of effective clinic scheme of noninvasivetemperature estimation using backscattered ultrasound, through data simulation, watertemperature experiments, in vitro tissue experiments, tissue characterization, clinicalresearch, and etc. It were found that the time-shift fitting order associated with theprior knowledge of heat mode and object size, the sampling frequency andsignal-noise ratio of the ultrasonic temperature measurement apparatus, tissuecharacterization parameter Km are the three major factors which directly affect thetemperature estimation precision. This thesis developed the clinical token method ofultrasound temperature estimation, considering the motion compensation, humanbody thermal model and anatomical model. It also gives the minimal ultrasonicsampling frame rate criterion, and the Km of the poker liver.更多还原