【作者】 盛磊；

【导师】 曾毅； 吴水才；

【作者基本信息】 北京工业大学 ， 生物医学工程， 2013， 博士

【摘要】 肿瘤热疗是指利用某种加热方法将肿瘤组织加热至超过其耐热温度以杀死癌细胞，治疗效果的实时无损评价是肿瘤热疗过程中一个亟需解决的问题。目前热疗监控成像主要有超声（US，Ultrasound）、核磁共振(MRI，Magnetic ResonanceImaging)、组织断层成像（CT，Computed Tomography）三种方式。CT由于对人体组织有损伤，MRI由于实时性较差、价格昂贵，且需要核磁兼容设备，都不便推广使用。超声因成本低、实时性好、与热疗设备兼容等优点，成为热疗监测的首选。本文研究了利用超声对微波热疗实时监测及相应系统设计。购置具有原始超声射频信号接口的高速超声数据采集器，搭建水浴加热实验及微波热疗实验平台，通过实验建立超声无损测温模型及超声组织定征模型，研究超声测温及生物组织热凝固区识别方法，并设计出相应微波热疗超声监测系统。主要研究内容如下：（1）实验平台的搭建。为了研究生物组织超声特征参数与温度的相关性，设计水浴实验系统。通过调节水箱温度，提取不同温度下的组织超声射频回波信号，建立超声无损测温及超声组织定征模型；为了利用超声监测微波热疗，设计微波热疗实验系统，提取热疗过程中组织超声射频回波信号，通过分析超声回波特征参数，识别热疗过程中的热凝固区；为了验证识别结果的准确性，统计微波热疗实验过程中不同加热功率和不同加热时间所获取的热凝固区大小。（2）生物组织超声特征参数的温度相关性研究。基于水浴实验平台，提取不同温度下的超声射频和视频特征参数，分析其与温度的相关性。其中超声回波射频信号参数主要包括超声背向散射积分、超声回波中心频率、超声衰减系数。研究结果表明：超声背向散射积分、超声衰减系数随温度升高而增大，且相关性系数大于0.95，超声回波中心频率随着温度的升高而降低。超声回波视频信号主要通过自建B超图像，分析图像的纹理特征参数与温度的相关性，分析了灰度直方图、灰度共生矩阵、灰度梯度共生矩阵等31个参数。研究结果表明：具有较高温度相关性的参数有灰度直方图中的灰度均值、灰度共生矩阵中的熵及灰度梯度共生矩阵中的混合熵、逆差距、相关。结合纹理参数的温度相关性分析和比值分析，可以得到随着组织的热凝固，超声纹理特征参数改变较大的有：灰度直方图统计量中的灰度均值，灰度梯度共生矩阵中的灰度不均匀性、能量、混合熵、惯性和逆差矩。微波热疗由于温度分布的不均匀性，用超声回波检测温度误差较大，很难达到临床应用要求。（3）组织微波热凝固区的超声无创检测研究。通过超声组织定征方法检测微波热疗过程中的热凝固区，并与实际测量的热凝固区大小对比，获取准确性最高的定征方法用于微波热疗监测系统的设计。采用的超声组织定征方法分别为超声背向散射积分、超声图像纹理特征参数、超声回波功率谱积分、散射元平均间距。研究结果表明：背向散射积分可以很好的检测出热凝固区，检测的速度较快，但由于组织的不均匀性，容易在热凝固区外获得识别结果；利用超声纹理特征参数进行热凝固区识别可以获得比较理想的结果，但由于声影的存在，检测结果只有实际消融区的一半，且纹理参数计算及神经网络的识别计算量很大，检测速度比较慢，不太适合热疗实时监测；利用超声功率谱积分值识别热凝固区，容易在热凝固区外检测到多余的组织，可能是利用频谱积分进行检测使用的数据太长，不易获得更加细节的信息；利用超声检测生物组织散射元平均间距的改变来检测热凝固区，既有较好的识别效果，又有较快的速度，可应用于肿瘤热疗的实时监测。（4）微波热疗超声无创监测系统的设计。通过MATLAB GUI实现肿瘤热疗超声无创监测系统的设计，利用微波热疗实验所获得的超声原始射频信号，重建B超图像，通过散射元平均间距的不同检测热凝固区，并将检测结果叠加到自建的B超图像之上，这样既能利用B超进行监测，在热疗过程中观察组织的整体状况，又能够通过灰度编码获取组织在热疗过程中的热凝固区变化。更多还原

【Abstract】 The tumor hyperthermia refers to the technique that heats tumor cells to exceedtheir heat resistance temperature to kill them. Real-time noninvasive evaluation oftreatment effect is a problem that needs to be resolved in microwave hyperthermia.Ultrasound (US), Magnetic Resonance Imaging (MRI) and Computed Tomography(CT) are the three main imaging methods available to noninvasively monitorhyperthermia therapy. However, as CT has damage to human tissues, and MRI haspoor real-time capability and high cost and requires nuclear magnetic compatibleequipment to operate, they are not widely used in hyperthermia monitoring. Becauseit is low cost, real-time and hyperthermia equipment compatible, ultrasound hasbecome the focus of the study. The real-time noninvasive monitoring system and itskey technologies for microwave hyperthermia were studied using ultrasoniccharacteristic parameters. The high-speed ultrasonic data collector with the originalradiofrequency (RF) signal interface was purchased. The water bath and microwaveheating experiment platforms were set up. The ultrasonic noninvasive temperatureestimation and tissue characterization models were established. The ultrasonictemperature estimation and tissue coagulation zone recognition methods were studied.The ultrasonic monitoring system for microwave hyperthermia was designed. Themain research refers to the following four parts:(1) Setting up of the experiment platform. The water bath heating experimentsystem was designed to study the correlations between the ultrasonic echo and thetemperature of biological tissues. The original ultrasound RF signal of tissues atdifferent temperature was acquired by adjusting the temperature of the water tank.The microwave heating experiment system was designed for ultrasonic monitoring ofmicrowave hyperthermia. The thermal coagulation zone was detected during thehyperthermia process by acquiring the original ultrasound RF signal of tissues andanalyzing the ultrasonic characteristic parameters of the echo. In order to verify theaccuracy of the detection results, the statistical analysis was conducted on the sizes ofthermal coagulation zones in the microwave heating experiment under differentheating powers and durations.(2) Studying of the correlations between ultrasonic characteristic parameters ofbiological tissues and temperature. Based on the water bath heating experimentplatform, the ultrasonic RF and video characteristic parameters at differenttemperature were extracted, and their correlations with temperature were analyzed.The characteristic parameters of ultrasonic echo RF signals included ultrasonicintegrated backscatter, center frequency of ultrasound echo, and ultrasonic attenuationcoefficient. Experimental results showed that the ultrasonic integrated backscatter and the ultrasonic attenuation coefficient increased with the rising of temperature, with thecorrelation coefficients being greater than0.95; the center frequency of ultrasoundecho decreased with the rising of temperature. The characteristic parameters ofultrasonic echo video signals were studied by self-reconstructing B-mode ultrasoundimages and analyzing the correlations between their texture features and temperature.A total of31texture feature parameters based on the gray level histogram, the graylevel co-occurrence matrices, and the gray level-gradient co-occurrence matrices wereanalyzed. Experimental results showed that the parameters with highly linearcorrelations with temperature included the mean gray scale of gray level histogram,the entropy of gray level co-occurrence matrix, and the hybrid entropy, inversedifference moment and correlation of gray level-gradient co-occurrence matrix.Combining correlation analysis and ratio analysis between texture parameters andtemperature, we found that with the thermal coagulation of tissues, the followingultrasound texture parameters took large changes: the mean gray scale of gray levelhistogram, and the intensity inhomogeneity, energy, hybrid entropy, inertia, andinverse difference moment of gray level-gradient co-occurrence matrix. Because ofthe inhomogeneous temperature distribution in the microwave heating experiment, theerror of temperature estimation by ultrasonic echo was big, so it was difficult to meetclinical application requirements.(3) Studying of the ultrasonic noninvasive detection of tissue coagulation zones inmicrowave hyperthermia. The thermal coagulation zones during the microwavehyperthermia process were detected using ultrasoic tissue chracterization techniques,and they were compared with the sizes of acutal measurement to get the detectiontechnique with the highest accuracy for the design of the microwave hyperthermiamonitoring system. The ultrasoic tissue characterization techniques were used basedon the ultrasonic integrated backscatter, ultrasound image texture features, ultrasoundecho power spectrum integration, and mean scatterer spacing. The ultrasonicintegrated backscatter can well detect the thermal coagulation zone with a highefficiency. However, due to the inhomogeneity of tissues, it tended to get recognitionresult outside the thermal coagulation zone, so the detected. Using the ultrasoundtexture features to identify thermal coagulation zones can obtain a satisfying result,but it would take a long time because of huge computation in calculation of textureparameters and the neural network based recognition. Besides, the detection resultwas only half of the real ablation zone due to the acoustic shadow. Consequently, theultrasound texture features were not suitable for real-time hyperthermia monitoring.The method of coagulation zone recognition using the ultrasound echo powerspectrum integration tended to detect redundant tissues outside the coagulation zone.The reason may be that the length of data used for detection by spectrum integration was too long, thus it was difficult to get more detailed information. Detecting thechanges of the mean scatterer spacing of biological tissues by ultrasound can be usedto recognize thermal coagulation zones, which can not only achieve better detectionresults, but also can be carried on in a high speed, so it can be applied in the real-timemonitoring of tumor hyperthermia.(4) Designing of the ultrasonic noninvasive monitoring system for microwavehyperthermia. The ultrasounic noninvasive monitoring systerm for tumorhyperthermia was designed using the MATLAB GUI. The original ultrasoic RFsignals from the microwave heating experiments were used to reconstruct B-modeultrasound images. The thermal coagulatoin zones were detected based on the changesof the mean scatterer spacing. The detection results were superimposed to thereconstructed B-mode images. The system can be used for B-mode ultrasound imagemonitoring and for observing the overal situation of tissues, and is capable to monitorthe changes of coagulation zones by grayscale coding during the hyperthermiaprocess.更多还原