【作者】 陈洪波；

【导师】 汤井田； 陈真诚；

【作者基本信息】 中南大学 ， 生物医学工程， 2009， 博士

【摘要】 治疗效果的实时无损评价是高强度聚焦超声(High IntensityFocused Ultrasound,HIFU)治疗中的一个急需解决的问题,也是一个集超声学、生物医学、计算机科学、智能信息处理、数字图像处理以及数学等众多领域在内的多学科交叉的前沿课题,有着巨大的应用需求。本论文针对当前HIFU治疗效果评价方法实时性不强的缺点,提出了利用超声成像方式对HIFU治疗效果进行实时评价。并重点研究了HIFU治疗效果实时评价中的三个关键问题,即:1)超声图像中斑纹噪声的抑制;2)HIFU辐射后焦域组织温度的无损测量;3)组织损伤情况的无损监控方法。对于斑纹噪声的抑制,本论文提出了采用独立分量分析方法去除同场景多幅图像中斑纹噪声的方法。同场景多幅图像中的图像信号和噪声信号被认为是相互独立的,独立分析方法则可以用来对这些图像进行分解,并将噪声信号从图像中得到有效地分离,以达到滤除图像中的斑纹噪声。实验结果表明,该方法能有效地滤除超声图像中的斑纹噪声,同时能较好地保留图像中的细节信息。对于无损测温,本论文利用实值离散Gabor变换对HIFU辐射前后的差图像进行变换后,能将反映组织温度变化的信息有效地提取出来。采用实值Gabor变换低频系数能量作为无损测温的手段,并利用离体猪骨骼肌和猪肝进行实验。实验证明,超声图像的实值离散Gabor系数能量和组织内温度之间的呈准线性关系,且具有较高的温度分辨率,比灰度平均值的无损测温能力要强,可以作为HIFU无损测温的方法。HIFU辐射后,组织损伤必然会引起B超图像的变化。对于监控组织损伤,本论文提出对HIFU焦域附近的差图像进行处理和分析,利用联合主分量分析和线性判别分析(combining Principal ComponentAnalysis with Linear Discriminant Analysis,PCA-LDA)方法和自组织映射(Self-Organizing Maps,SOM)方法分别对差图像进行特征提取,以识别HIFU辐射后焦域组织是否已经受到损伤。在对离体猪骨骼肌组织和离体猪肝组织进行了组织损伤识别实验。实验结果比较分析认为,SOM方法能得到最好的识别率,对于猪骨骼肌组织样本识别率高达96.1%,对于猪肝组织样本识别率高达91.7%。PCA-LDA方法的识别要低于SOM方法,这是因为PCA-LDA方法中的分类规则采用的是线性分类,而由SOM方法的聚类结果可以看到,组织损伤的特征并不是绝对线性可分的。对HIFU治疗效果实时评价中的.临床方案做了总体设计,并讨论了在临床应用中需要考虑的一些问题,如因HIFU辐射能量分布不均匀导致的焦域的温度场不均匀的问题,数据库的建立问题,HIFU换能器和超声探头的位移的问题以及HIFU辐射参数的确定问题等。更多还原

【Abstract】 Noninvasive Estimation of Treatment Effect (NETE) is a problem that needs to be resolved in High Intensity Focused Ultrasound (HIFU). It is an advanced interdisciplinary topic which involves ultrasonics, biomedicine, computer science, intelligence information processing, digital image processing, mathematics, etc. It has a great application requirement.Since current schemes of NETE are difficult to estimate the treatment effect in real-time, a scheme that is based on ultrasound image is proposed for NETE. Three main problems in NETE have been studied. They are: 1) reducing the speckle noise in ultrasound image; 2) noninvasive temperature estimation for HIFU; 3) noninvasive detection of lesion in tissue.In order to reduce speckle noise in multiple copy ultrasound images, a method that is based on independent component analysis is proposed. The original image signal and speckle noise signal are separated by this method after supposing that the original image and noise are mutually independent in multiple copy images. Experimental results show that the proposed method can restrain effectively the speckle noise in ultrasound image.In tissue, temperature changes induced by HIFU exposures result in gray variation of the B-mode ultrasonic image that can be used to estimate temperature. Thus, a method is proposed to noninvasively estimate temperature based on Real-valued Discrete Gabor Transform (RDGT). The difference image is decomposed to two types of information using this method: the information that is correlated to temperature, and the other information, such as noise. In-vitro fresh pork and liver tissue are used in experiment. The experimental results show that temperature is approximately linear to energy of RDGT, and the new method has a high temperature resolution. The RDGT can be applied to estimate temperature for HIFU. Lesion in tissue, created by HIFU exposures, inevitably leads to some new interfaces in tissue and changes in B-mode ultrasound image. Feature of tissue lesion in difference-sub-image is extracted using Self-Organizing Maps (SOM) and the method combining Principal Component Analysis with Linear Discriminant Analysis (PCA-LDA), respectively. Fresh pork and liver in vitro are used in experiments. The experimental results show that the SOM method can get the best result for lesion recognition, the recognition rate of pork tissue and liver tissue are 96.1% and 91.7%, respectively. The recognition rate for PCA-LDA method is lower than that of SOM method because the feature of tissue lesion is linear indivisible.An effective clinic scheme is developed for NETE, and some problems are discussed. For example, nonuniform temperature field conduced by nonuniform energy field; tissue database construction; movement of HIFU transducer with ultrasonic transducer; rectification of HIFU parameters.更多还原